Rust Forge

Welcome to the Rust Forge! Rust Forge serves as a repository of supplementary documentation useful for members of The Rust Programming Language. If you find any mistakes, typos, or want to add to the Rust Forge, feel free to file an issue or PR on the Rust Forge GitHub.

Help Wanted

Want to contribute to Rust, but don't know where to start? Here's a list of rust-lang projects that have marked issues that need help and issues that are good first issues.

Current Release Versions

See the release process documentation for details on what happens in the days leading up to a release.

No Tools Breakage Week

To ensure the beta release includes all the tools, no tool breakages are allowed in the week before the beta cutoff (except for nightly-only tools).

External Links

• Bibliography of research papers and other projects that influenced Rust.
• Rust Pontoon is a translation management system used to localize the Rust website.

Platforms

Rust uses a number of different platforms for organizing work and internal communications between teams. This does not currently seek to be an exhaustive list, rather documenting the policies for a select few platforms used by the teams.

Twitter

The Rust project has a number of official Twitter accounts, credentials for which are currently maintained by the infrastructure team.

• @rustlang
• @RustStatus
• @cratesiostatus
• @rustasync

Twitter Guidelines

The project runs the Twitter account [@rustlang] The account is handled by a small team of volunteers.

The account will mostly tweet links to the Rust blog and Rust Insiders blog. Additionally it will retweet:

• links to blog posts about Rust, retweeting the original author if possible
• questions about Rust, so all followers can help
• Meetup or conference announcements
• announcements of new Rust projects
• anything else relevant

We will not retweet:

• content that bashes other programming languages/projects or is otherwise unconstructive in its discussion of language/tech choice
• Personal announcements ("Today I start my job at $COMPANY writing Rust")
• Learning Rust updates ("Today I started to learn Rust")

The Direct Messages are open to everyone. If someone wants something retweeted, they should send the tweet via DM. The vast majority of these things should be retweeted, keeping it to the above rules. Requests of an author via DM or Tweet to not retweet something will be honored.

Additionally account handlers may look through the #rustlang hashtag for noteworthy content.

The account will only follow a small number of Project-owned/related Twitter accounts. At the time of writing (February 2022) this is only @cratesiostatus and @rust_foundation.

Access

Currently access to all four accounts is granted together via a 1password vault; we don't split this into more fine-grained access. Some automation uses API keys of the status accounts to automatically tweet about upcoming events on crates.io.

Access is limited to a small set of folks in the twitter marker team; this isn't automated (changes should ping infra admins for provisioning access).

People with access to 1password should:

• Never change the password or take other administrative action (this is only to be done by infra admins)
• Exclusively use the project-hosted instance to keep a copy of the password (don't save it to any other password database, including in browser)
• Never share the password with others (even if they're in the list)
  • All access should always go through regular channels to ensure we're not accidentally leaking the password by passing it through unsecure channels (e.g., email)
• Be aware that the password may change regularly (requiring re-authorization)

If you believe you should have access, please file a PR against the team repository requesting it and note in the description that you've read this policy.

Discord

Rust's Discord is currently used by a variety of teams such as Community, Ops, and Documentation, as well as their working groups. It is also maintained as a communication tool for Domain Working Groups, and provides a space for general discussion among Rust users, contributors, and beginners.

Where to go for help with using Discord

Discord's support center provides documentation about its user interface and account settings.

Getting started

1. Understand community standards
   Discord, like all official Rust spaces, is governed by the Code of Conduct. Before joining the conversation there, you can prepare by reading the Code of Conduct and Moderation Guidelines. It is also useful to read Discord's Community Guidelines

2. Access channels
   To access the Rust Discord, visit https://discord.gg/rust-lang. If you do not already have a Discord account, you can register for one as part of the process of gaining access. Your first action should be agreeing to our Code of Conduct by following the instructions in #welcome.

3. Configure notifications
   It is a good idea to disable notifications for channels that are not relevant to you, so that you will not be overwhelmed with messages. Select the expansion arrow next to the server name banner (titled "The Rust Programming Language") and select Notifications from the dropdown. Then follow the configuration instructions provided on the Discord Support site.

Appropriate conversation

Discussions should be related to the channel purpose. On team channels, conversation should be related to team business. All channels are expected to be used for purposes related to the Rust project. Discussion of (for example) wildlife or sightseeing are not appropriate.

Channels

The following channels are relevant to newcomers to the Rust project:

• welcome - Where you agree to the CoC.
• rust-usage - This is a channel where you can access support for resolving specific language use questions. The Rust Users Forum is also relevant to your needs.
• beginners - Here, you can meet people who began using Rust relatively recently.
• contribute - Interested in contributing to the Rust project? In addition to joining this channel, you can subscribe to the This Week In Rust newsletter, where many opportunities are regularly posted. It may also help to find out more about specific teams.

Channels outside of General are for contributors to Rust.

Messages

Discord conversation takes place when people are available, so you should not generally expect that your messages will receive a response quickly unless a meeting is taking place. Depending on how your notifications are configured, you will see a red circle on top of the Discord icon in your system tray when new messages are received. If you wish to communicate with a specific individual, right-click on their user icon and select "Message" in the dropdown menu.

Read-only view

Set up a Discord account (as described in Getting Started, above) in order to access Discord. There is not currently a read-only archive view available.

Email

While most of Rust's discussion happens on other platforms, email is eternal and we occasionally need a way to approach individuals or groups privately. Our email is hosted through Mailgun (provided by Mozilla). We create and edit the mailing lists for teams through the rust-lang/team repository. Our email domain is rust-lang.org, e.g. ferris@rust-lang.org.

Sending a public broadcast

If your teams need to reach everyone in the Rust organisation, they can send an email to all@. It is recommended that you only use this mailing list when you know that you need to contact every member, such as for organising a members event like the All Hands, or for security alerts.

Keeping responses private

When sending a message to all@, do not put all@ in To. This will mean that any replies to your broadcast will also be sent to everyone. Instead, put your team's email address in To field, and place all@ in the Bcc field. Then any replies will be sent to just your team.

GitHub

Github is where the Rust project hosts all of its code, as well as large parts of its discussions.

Organisations

• rust-lang — The Rust project organisation.
• rust-embedded — The Embedded Working Group organisation.
• rustwasm — The WebAssembly Working Group organisation.
• rust-cli — The Command Line Application Working Group organisation.
• rust-secure-code — The Secure Code Working Group organisation.
• rust-gamedev — The Game Development Working Group organisation.

Team Repositories

• Compiler Team
• Core Team
• Crates.io Team
• Infrastructure Team
• Language Team
• Moderation Team
• Release Team

Administration FAQ

Who administrates the rust-lang organisation?

All core team members have the admin role.

How do I create a new repository under the rust-lang organisation, or make changes that require admin level permissions?

You can contact a GitHub admin directly, or post the request in project leads (public) stream on Zulip and an admin will be able to respond to your request there.

Zulip

Rust's Zulip is used by a number of teams, notably the compiler, language, and library teams, along with their working groups.

Zulip can be an unintuitive platform to get started with. To get started, take a look at the getting started guide. For more detail, examine the Zulip user documentation!

Where to go for help with using Zulip

If you're testing a feature, or want to get help, the #zulip stream is the place to go. Like elsewhere, the best thing to do is to create a new topic for each question.

Getting started

It is recommended to first look at the official getting started guide. Like Rust itself, Zulip is a bit special and reading the documentation before digging can be really helpful.

You'll definitely want to configure the streams that you're subscribed to when getting started; the default set is quite limited, and there are many groups that exist beyond it. Subscribing to a stream is very low cost -- it is similar to being "in" an IRC channel, except that logs are available for all streams, regardless of subscription status.

It's not necessary to introduce yourself, but feel free to say hello in the #new members stream.

User groups

User groups can be pinged by anyone with the @<group> notation, same as pinging another user. Groups can be created by anyone, and anyone can join a group.

Users should feel free to join (or leave) groups on their own. Furthermore, users should feel free to create groups as needed, though it is currently expected that this is somewhat rare. You should name your group similar to how you would name a stream for the same purpose, though groups can be more fine-grained (or less). For example, @T-compiler/meeting currently does not have a dedicated stream.

Appropriate conversation

In most streams, you should try to keep conversations related to team business. The #general stream is a bit broader, but even there, discussions should be closely related to Rust (though may not relate to projects of any particular team). All channels are expected to be used for discussions related to the Rust project, though; discussions of (for example) wildlife or sightseeing are not appropriate.

Streams

These are similar to "channels" on other platforms (i.e., there should not be too many). On the other hand, you can choose which streams you subscribe to, so there can be more than channels on other platforms. Read Zulip's documentation for more details.

Streams are appropriate for any Rust official group. For example, working groups, project groups, teams are all examples of official groups. These should ideally also be represented in the team repository.

Default streams

This section is still under debate, and it is not yet clear which direction we will go. It is non-normative, and should not be used yet for modifications to the Zulip instance.

The default set of streams is chosen to allow incoming people to be able to have at least one place to go that can then, if necessary, direct them to a more specific location.

Currently that means that every top-level group present on Zulip is by default visible. Specifically, no stream that contains a / will be enabled by default.

Currently this set is:

• general
• t-lang
• t-compiler
• t-libs
• project-ffi-unwind
• project-inline-asm
• project-safe-transmute
• rust-survey-2019
• wg-async-foundations
• wg-database
• wg-formal-methods
• wg-secure-code
• wg-traits
• zulip

An alternative, minimalistic, approach is to use:

• general
• zulip
• announce
• new members

as the default set, which would push people into customizing their default set when starting out.

Stream naming

A stream should be named such as #t-{team}/{group name}. For example, #t-compiler/wg-parallel-rustc. More levels of nesting are fine, e.g., a working group might want "subgroups" as well, though you may want to omit the team name in such a case -- keeping the stream name short is good for usability, to avoid confusion between different streams which share the same prefix.

If no top-level team exists, or the group spans multiple teams (e.g., project-ffi-unwind), then the top level team should be omitted.

Streams should be clearly communicated as being for a specific purpose. That purpose can be broad, but it should likely include a group of some kind (even if that group is transient, e.g., people who are having trouble with the rust build system, or people working on the compiler). Furthermore, we do not currently intend for this Zulip to be a general place for community projects not affiliated with the Rust organization; if they wish to use Zulip, it is free for open source.

When a new stream is created, you should announce it in #announce. This is generally done automatically by Zulip.

Topics

A topic is attached to every message within a given stream (these are the subdivisions within streams). Topics are generally transient, and live for as long as there is active discussion on a topic. Thinking of topics like email subjects is helpful.

New conversation in a given stream should almost always start in a new topic, not a preexisting one. Unlike (for example) GitHub issues, you should not attempt to search for a past topic on the same subject. Do not spend too long on the name of the topic, either, beyond trying to make it short. Topics should generally be no longer than 20 characters (loosely two to three words), to make sure it is visible to users.

You should eagerly fork new discussion topics into fresh topics. Note that this can be done with the tail of another topic (if accidentally you diverge into another area of discussion).

To fork from an existing topic, see Zulip's documentation here.

Messages

Zulip is a unique platform which combines synchronous and asynchronous communication in one location. You should not generally expect that your messages will receive a response quickly, and unlike (for example) Discord, there is likely not much reason to "re-ping" on a particular issue every few hours as your message is unlikely to vanish into history, being isolated to a specific topic.

Linkifiers

Our Zulip supports a lot of helpful linkifiers, and we're generally happy to add more on request. See the documentation for the format. Propose one in #zulip!

Generally, github-org/repo#123 works for linking to an issue or PR; the below list gives a few more "special cased" repositories.

Don't forget that standard Markdown syntax for links also works.

We currently support linking to issues on a few repositories:

• rust-lang/rust with #4545 or rust#4545
• rust-lang/rfcs with RFC#3434 or rfc#3434
• rust-lang/async-book with async-book#2334
• rust-lang/chalk with chalk#2334
• rust-lang/compiler-team with compiler-team#3433
• rust-lang/ena with ena#3434
• rust-lang/miri with miri#3434
• rust-lang/polonius with polonius#3434
• rust-analyzer/rust-analyzer with rust-analyzer#3434
• rust-lang/rustc-dev-guide with rustc-dev-guide#3434
• rust-lang/stdarch with stdarch#3434
• rust-lang/team with team#3434
• rust-lang/unsafe-code-guidelines with ucg#3434

We currently support linking to commits on these repositories:

• rust-lang/rust with 40-character long SHAs, e.g., 25434f898b499876203a3b95c1b38bad5ed2cc5d

Read-only view

Our Zulip instance has the web-public streams beta feature enabled, and we use it for all public streams. Please let us or Zulip developers know if there's any problems with this. The previous solution to the web-public view was the zulip archive, which now redirects to the web public view.

Zulip Moderation

Zulip, like all official Rust spaces, is governed by the Code of Conduct. If you have concerns, please feel free to escalate to the moderation team.

However, though the moderation team is the top-level body here, it is not the only place where you can seek help with moderation within Zulip.

One method for reaching the Zulip administrators privately is to email zulip-admin.239bd484c0347d2d43214d8581f3e125.show-sender@streams.zulipchat.com. See this page for details on how this works.

You can also ping the @mods group on Zulip; note that this will be public.

It is not currently possible for normal users to self-administrate (e.g., muting another user). However, each individual stream, including private streams, can be muted:

For admins/moderators

Some common actions for moderators are listed on this page.

Notably,

• in "Organization permissions" we can restrict users to mandate invitations before joining (this is the "no new users" button)

New admins/moderators should add themselves to the mods group on Zulip. (Note that this is something that any user can do!)

Triagebot

Triaging on the rust-lang repository is an important step to take care of issues. The triagebot is the tool that allows anyone to help by assigning, self-assigning or labeling issues without being a member of the rust-lang organization.

To enable triagebot on a particular repository (currently only in the rust-lang organization), add a triagebot.toml file in the repository root. It should have a section per "feature". Please read this page to learn how to enable each feature and the options supported; if you spot something missing please let us know by filing an issue, thanks!

• Issue assignment
• Issue notifications
• Ping a team
• Glacier
• Triage
• Apply labels to issues
• Request prioritization
• Autolabel an issue
• Notify Zulip
• Major Changes

Issue assignment

Any user belonging to the rust-lang organization can claim an issue via @rustbot claim or if the user is not part of the rust-lang organization rustbot will assign the issue to itself; then it will add a "claimed" message in the top-level comment, to signal who the current assignee is. It is possible to override someone else's claim (no warning/error is given).

You can drop your claim to the issue via @rustbot release-assignment; Rust team members can do the same if they want to release someone else's assignment.

@rustbot assign @user can be used only by Rust team members and will assign that user to the issue (with same rules as before -- either directly or indirectly).

Soon (when the "highfive" bot migration will be complete, see rust-lang/highfive#258), r? will also assign reviewers to PRs, though unlike issues, non-team members cannot be assigned. Anyone can invoke the command.

To enable on a repository, add the following to a triagebot.toml file in the repository root.

[assign]

Issue notifications

Each registered team member has a notifications page at:

https://triage.rust-lang.org/notifications?user=<github-username>

This page is populated from direct mentions (@user) and team mentions (@rust-lang/libs) across the rust-lang organization.

It can also be edited via Zulip by private-messaging triagebot. Any Rust organization member can edit their notifications page, or pages of other Rust organization team members. To do so, the editor must have a zulip-id listed in their people/username.toml file in the team repository. The bot will tell you which ID to use when talking to it for the first time; please r? @Mark-Simulacrum on PRs adding Zulip IDs.

The following commands are supported:

• acknowledge <url> (or short form ack <url>)
• acknowledge <idx> (or short form ack <idx>)

These both acknowledge (and remove) a notification from the list.

• acknowledge all or acknowledge * (or short form ack all or ack *)

This acknowledges and removes all notifications.

• add <url> <description... (multiple words)>

This adds a new notification to the list.

• move <from> <to>

This moves the notification at index from to the index to.

• meta <idx> <metadata...>

This adds some text as a sub-bullet to the notification at idx. If the metadata is empty, the text is removed.

• as <github username> <command...>

This executes any of the above commands as if you were the other GH user.

Ping a team

The bot can be used to "ping" teams of people that do not have corresponding Github teams. This is useful because sometimes we want to keep groups of people that we can notify but we don't want to add all the members in those groups to the Github org, as that would imply that they are members of the Rust team (for example, Github would decorate their names with "member" and so forth). The compiler team uses this feature to reach the notification groups.

When a team is pinged, we will both post a message to the issue and add a label. The message will include a cc line that @-mentions all members of the team.

Teams that can be pinged

To be pinged, teams have to be created in the Rust team repository. Frequently those teams will be marked as marker-team, meaning that they do not appear on the website. The LLVM team is an example.

Configuration

To enable the team (e.g. TeamName) to be pinged, you have to add section to the triagebot.toml file at the root of a repository, like so:

[ping.TeamName]
message = """\
Put your message here. It will be added as a Github comment,
so it can include Markdown and other markup.
"""
label = "help wanted"

This configuration would post the given message and also add the label help wanted to the issue.

You can also define aliases to add additional labels to refer to same target team. Aliases can be useful to add mnemonic labels or accomodate slight mispellings (such as "llvms" instead "llvm"), see the following example:

[ping.cleanup-crew]
alias = ["cleanup", "cleanups", "shrink", "reduce", "bisect"]
message = """\
message content...
"""

This will allow the command @rustbot ping cleanup-crew to be understood with all the aliased variants, ex.:

@rustbot ping cleanup
@rustbot ping shrink
...

Check out the rust-lang/rust configuration for an up-to-date examples.

Pinging teams

To ping the team XXX, simply leave a comment with the command:

@rustbot ping XXX

Related issues

• Requested in https://github.com/rust-lang/triagebot/issues/169

Glacier

This adds the option to track ICEs (Internal Compiler Errors). Do note that the GitHub Gist must be from a Rust Playground link. The link must also be in quotes (""), example:

@rustbot glacier "https://gist.github.com/rust-play/xxx"

where xxx is the SHA1 hash of the GitHub gist generated by the Playground "share" button.

Triage

This command can be used by people in charge of prioritizing issues, to assign either low or high priorities to issues. This is mostly done by the Compiler Prioritization WG for compiler bugs.

@rustbot triage {high,medium,low}

The configuration for this feature is:

[triage]
remove = ["I-prioritize"] # the set of labels to remove when this command is invoked
high = "P-high"
medium = "P-medium"
low = "P-low"

Apply labels to issues

This command lets anyone apply labels to issues. This is most useful when opening an issue. In general, labels get applied to issues by the Triage WG. If you are interested in helping triaging issues, see the Triage WG procedure.

The specific grammar can be found here, but some examples are listed below. The grammar is intended to be fairly intuitive for people, to prevent needing to reach for documentation when using the bot.

@rustbot modify labels to +T-lang, -T-compiler

This will remove the T-compiler label and add the T-lang label. You can also omit the + sign, if you want, and it'll be implied.

You can also write the same command in a few other ways:

@rustbot modify labels to +T-lang and -T-compiler
@rustbot modify labels: +T-lang and -T-compiler
@rustbot modify labels to +T-lang -T-compiler

Note that the command can either terminate with a . or a newline, otherwise the bot will not parse the command successfully.

Errors

The bot currently restricts the labels that can be applied by people outside the Rust teams. For example, they can't add the I-unsound label. Most of the time, you shouldn't hit this. Feel free to ping the release team if you feel that a label should be added to the set of allowed labels!

Enabling

[relabel]
# any label is allowed to be set by team members (anyone on a team in rust-lang/team)
# but these can be set by anyone in the world
allow-unauthenticated = [
    "C-*", # any C- prefixed label will be allowed for anyone
           # independent of authorization with rust-lang/team
    "!C-bug", # but not C-bug (order does not matter)
]

Request prioritization

Users can request an issue to be prioritized by the Prioritization WG.

To do so, you can invoke the following command:

@rustbot prioritize

This will simply add the I-prioritize label to the issue.

Errors

The command fails if the issue has already been requested for prioritization (i.e. already has the I-prioritize label).

Enabling

[prioritize]
# Name of the label used for requesting prioritization on issues
label = "I-prioritize"

Autolabel an issue

When certain labels are added to an issue, this command will trigger adding a set of additional prioritization labels to the issue. In the following example adding the "I-prioritize" label will automatically add the labels in trigger_labels but only if the issue is not already labeled with those in exclude_labels (this is to avoid applying unrelated labels to issues).

[autolabel."I-prioritize"]
trigger_labels = [
    "regression-from-stable-to-stable",
    "regression-from-stable-to-beta",
    "regression-from-stable-to-nightly"
]
exclude_labels = [
    "P-*",
    "T-infra",
    "T-release"
]

Notify Zulip

When a prioritization label is added to an issue, this command will create a new topic on Zulip, in the designated stream ("245100" in the following example), replacing {number} and {title} with the issue GitHub ID and title:

[notify-zulip."I-prioritize"]
zulip_stream = 245100 # t-compiler/wg-prioritization/alerts
topic = "I-prioritize #{number} {title}"
message_on_add = "@**WG-prioritization** issue #{number} has been requested for prioritization."
message_on_remove = "Issue #{number}'s prioritization request has been removed."

The subscribers of that Zulip stream will receive a notification and can discuss the prioritization of the issue.

Major Changes

A major change is an issue that will have a big impact on users. See this page on the MCP process for detailed explanations.

The compiler team uses the major change process, which requires:

• An issue
• A "second", who is an expert from the Compiler team who thinks the proposal is a good idea

We have supporting automation for both parts.

First, on the rust-lang/compiler-team repository, an issue with the "major-change" label (MCP = Major Change Proposal) is created via the template. Once that's opened, it automatically gains the "to-announce" label which should be removed when it's announced at a compiler team meeting.

For seconds, you tell rustbot @rustbot seconded or @rustbot second and it will apply the relevant label. Only team members can do so.

Configuration:

[major-change]
# Label to apply once an MCP is seconded
second_label = "final-comment-period"
# Label to apply when an MCP is created
meeting_label = "to-announce"
# The Zulip stream to automatically create topics about MCPs in
# Can be found by looking for the first number in URLs, e.g. https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler
zulip_stream = 131828

Core

This section documents policies established by the core team. These policies tend to apply for "project-wide resources", such as the Rust blogs.

Rust Blog Guidelines

Context

The Rust project maintains two blogs. The “main blog” (blog.rust-lang.org) and a “team blog” (blog.rust-lang.org/inside-rust). This document provides the guidelines for what it takes to write a post for each of those blogs, as well as how to propose a post and to choose which blog is most appropriate.

How to select the right blog: audience

So you want to write a Rust blog post, and you’d like to know which blog you should post it on. Ultimately, there are three options:

• The main Rust blog
  • Suitable when your audience is “all Rust users or potential users”
  • Written from an “official position”, even if signed by an individual
• The team Rust blog
  • Suitable when your audience is “all Rust contributors or potential contributors”
  • Written from an “official position”, even if signed by an individual
• Your own personal blog
  • Everything else

There are two key questions to answer in deciding which of these seems right:

• Are you speaking in an “official capacity” or as a “private citizen”?
• Who is the audience for your post?

In general, if you are speaking as a “private citizen”, then you are probably best off writing on your own personal blog.

If, however, you are writing in an official capacity, then one of the Rust blogs would be a good fit. Note that this doesn’t mean you can’t write as an individual. Plenty of the posts on Rust’s blog are signed by individuals, and, in fact, that is the preferred option. However, those posts are typically documenting the official position of a team — a good example is Aaron Turon’s classic post on Rust’s language ergonomics initiative. Sometimes, the posts are describing an exciting project, but again in a way that represents the project as a whole (e.g., Manish Goregaokar’s report on Fearless Concurrency in Firefox Quantum).

To decide between the main blog and the team blog, the question to ask yourself is who is the audience for your post. Posts on the main blog should be targeting all Rust users or potential users — they tend to be lighter on technical detail, and written without requiring as much context. Posts on the team blog can assume a lot more context and familiarity with Rust.

Writing for the Main Rust blog

The core team ultimately decides what to post on the main Rust blog.

Post proposals describing exciting developments from within the Rust org are welcome, as well as posts that describe exciting applications of Rust. We do not generally do “promotional cross-posting” with other projects, however.

If you would like to propose a blog post for the main blog, please reach out to a core team member. It is not suggested to just open PRs against the main Rust blog that add posts without first discussing it with a core team member.

Release note blog posts

One special case are the regular release note posts that accompany every Rust release. These are managed by the release team and go on the main blog.

The blog posts are published on the same day as the release by the same person in the release team running the release. Releases always happen on Thursdays.

Before publishing a release post, it goes through a drafting process:

1. The milestone (e.g. for 1.39.0) for the release is consulted.
2. PRs that we think are sufficiently important are included, and some items are headlined. The writing of a blog post typically happens through a hackmd document.
3. Headlined items are sometimes written by different people, and we try to peer-review each subsection.
4. The blog post draft is submitted as a PR on the blog repo for final review a few days before the release.

Team Rust blogs

Teams can generally decide for themselves what to write on the team Rust blog.

Typical subjects for team Rust blog posts include:

• New initiatives and calls for participation
• Updates and status reports from ongoing work
• Design notes

To propose a blog post for the team blog of a particular team, reach out to the team lead or some other team representative.

Community

This section documents the processes of the community team, and related projects.

External Links

• The Community team GitHub repository contains information about how the community team organizes.
• The RustBridge website contains information on hosting your own local RustBridge event.
• Rustlings is an project with small exercises designed around getting newcomers used to reading and writing Rust.

State of Rust Survey FAQ

In this FAQ we try to answer common questions about the Annual State of the Rust Language Community Survey. If in your opinion there is a missing question or if you have a concern about this document, please do not hesitate to contact the Rust Community Team or open an issue with the Community Team.

Why is this survey important for the Rust project?

Rust is an Open Source project. As such, we want to hear both from people inside and outside our ecosystem about the language, how it is perceived, and how we can make the language more accessible and our community more welcoming. This feedback will give our community the opportunity to participate on shaping the future of the project. We want to focus in the requirements of the language current and potential users to offer a compelling tool for them to solve real world problems in a safe, efficient and modern way.

What are the goals of the survey?

• To understand the community's main development priorities and needs
• To categorize the population of users of the language
• To focus our efforts on events and conferences to drive more impact
• To identify potential new contributors to the community goals

How much time will it take to answer the survey?

In average, it should take from 10 to 15 minutes.

What kind of questions are included in the survey?

It includes some basic questions about how do responders use Rust, their opinion the ecosystem's tools and libraries, some basic questions regarding the responders' employer or organization and their intention to use Rust, technical background and demographic questions and some feedback related to the Rust project's community activities and general priorities.

How will we use the data from the survey responses?

The answers from the survey will be anonymized, aggregated, and summarized. A high level writeup will be posted to https://blog.rust-lang.org.

How is personally identifiable information handled?

Nearly every question in the survey is optional. You are welcome to share as much or as little information as you are comfortable with. Only the Rust language Core Team and the Community Team Survey Leads will have access to the raw data from the survey. All the answers are anonymized prior to be shared with the rest of the teams and prior to the results publication.

Why is the survey collecting contact information?

The survey optionally collects contact information for the following cases if you expressed interest in:

• future conferences or meetups in your area
• helping to organize a Rust event, meetup, or conference
• talking to a Rust team about using Rust inside your company
• Rust training
• interest in a Rust team contacting you about your survey responses

If you would like to be contacted about any of this, or any other concerns, but you don't want to associate your email with your survey responses, you can instead email the Rust Community Team at community-team@rust-lang.org or the Core Team at core-team@rust-lang.org, and we will connect you to the right people.

Where and when is the survey results report published?

We expect to publish results from the survey within a month or two of the survey completion. The survey results will be posted to project's blog.

Where can I see the previous survey reports?

• State of Rust 2018
• State of Rust 2017
• State of Rust 2016

Compiler

This section documents the Rust compiler itself, its APIs, and how to contribute and provide bug fixes for the compiler.

External Links

• The Rustc Dev Guide documents how the compiler works as well providing helpful information to help get new contributors involved in the development.
• Rustc's internal documentation.
• The Compiler team website is the home for all of the compiler team's planning.
• oli-obk's FIXME page lists all of the FIXME comments in the Rust compiler.

Cross Compilation

This subsection documents cross compiling your code on one platform to another.

Windows

1. Acquire LLD somehow. Either your distro provides it or you have to build it from source.
2. You'll need an lld-link wrapper, which is just lld using the link flavor so it accepts the same flags as link.exe. You may either have a binary called lld-link, or you may have to write some sort of script to wrap lld.
3. If you want to be able to cross compile C/C++ as well, you will need to obtain clang-cl, which is clang pretending to be cl.
4. You'll need libraries from an existing msvc installation on Windows to link your Rust code against. You'll need the VC++ libraries from either VS 2015 or VS 2017, and the system libraries from either the Windows 8.1 or Windows 10 SDK. Here are some approximate paths which may vary depending on the exact version you have installed. Copy them over to your non-windows machine.
   • VS 2015: C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\lib
   • VS 2017: C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Tools\MSVC\14.10.24728\lib
   • Windows 10 SDK: C:\Program Files (x86)\Windows Kits\10\Lib\10.0.14393.0
   • Windows 8.1 SDK: C:\Program Files (x86)\Windows Kits\8.1\Lib\winv6.3
5. If you want to cross compile C/C++ you'll also need headers. Replace lib in the above paths with include to get the appropriate headers.
6. Set your LIB and INCLUDE environment variables to semicolon separated lists of all the relevant directories for the correct architecture.
7. In your .cargo/config add [target.x86_64-pc-windows-msvc] linker = "lld-link" or whatever your lld pretending to be link.exe is called.
8. For cross compiling C/C++, you'll need to get the gcc crate working correctly. I never tested it to cross compile, I have no idea whether it will even do anything sane.
9. Install the appropriate target using rustup and pass --target=x86_64-pc-windows-msvc while building. Hopefully it works. If it doesn't, well... I don't know.

Cross-team Collaboration

If you are a member of another team and would like to raise an issue with the compiler team..

..for discussion

Write a comment on a GitHub issue describing the reason for the nomination (i.e. what decision needs to be made/what opinion is sought; what are the relevant parts to the compiler team, etc) and add the I-compiler-nominated label to a issue (you can include @rustbot label +I-compiler-nominated in your comment to do this).

Once nominated, the issue will be discussed in a upcoming triage meeting. The compiler team doesn't always get through all nominated issues each week, so it can take more than one meeting for your issue to be discussed.

Once discussed, a member of the team will comment on the issue with the conclusion of the discussion and linking to the relevant Zulip chat.

..to be fixed

If there is an existing working relationship between a member of the requesting team and a contributor to the compiler, then the first option that a team has for requesting tasks be completed is to ping that contributor and ask if they can complete the task. It is recommended that pings take place in public Zulip channels so that..

• ..other contributors that have free time have the opportunity to offer their help.
• ..other compiler team members/leadership can ensure that requests being made are reasonable (see the rest of this section for the types of issues that the compiler team commits to prioritizing on behalf of other teams).

It is worth considering the available bandwidth of the contributor that the request is being made of, and whether their areas of expertise in the compiler are relevant.

When there is not a appropriate contact in the compiler team to reach out to directly, write a comment on a GitHub issue (or create an issue) describing the task that needs completed. Teams should nominate issues for the compiler team when issues..

• ..are not already tracked by/part of an existing initiative or working group and..
• ..are blocking/impeding the work of the other team (e.g. a feature or bug preventing the stabilization of something otherwise complete), but..
• ..aren't absolutely mission-critical - a soundness bug or otherwise critical issue will be prioritized by the prioritization working group and addressed through the compiler team's other processes for these bugs. If the issue lacks a prioritization label, you can add the I-prioritize label and it will be enqueued for prioritization.

A detailed description of the feature being requested or the bug to be fixed is helpful wherever possible (so that the compiler contributor does not need to make a guess as to a solution that would solve the problem for the requesting team). If a member of the requesting team isn't explicitly listed as the point-of-contact for the issue, then the author of the comment will be assumed to be the point-of-contact.

Add the I-compiler-nominated label to a issue (you can use @rustbot label +I-compiler-nominated to do this).

Once nominated, the issue will be discussed in a upcoming triage meeting. The compiler team doesn't always get through all nominated issues each week, so it can take more than one meeting for your issue to be discussed. In the compiler team's discussion, the issue may..

• ..be accepted, in which case it will be assigned to a contributor and the nomination label removed. Once assigned, a member of the team will work on the issue. If no work is completed after a reasonable time, then re-nominate the issue and the compiler team will find someone else to complete the work.
• ..or not accepted (e.g. due to insufficient bandwidth, other critical/high-priority bugs, being unable to find an appropriate contributor, or the issue lacking feasibility). In this case, the compiler team will reply to the nomination with an explanation and will remove the nomination label.

Review policies

Every PR that lands in the compiler and its associated crates must be reviewed by at least one person who is knowledgeable with the code in question.

When a PR is opened, you can request a reviewer by including r? @username in the PR description. If you don't do so, rustbot will automatically assign someone.

It is common to leave a r? @username comment at some later point to request review from someone else. This will also reassign the PR.

bors

We never merge PRs directly. Instead, we use bors. A qualified reviewer with bors privileges (e.g., a compiler contributor) will leave a comment like @bors r+. This indicates that they approve the PR.

People with bors privileges may also leave a @bors r=username command. This indicates that the PR was already approved by @username. This is commonly done after rebasing.

Finally, in some cases, PRs can be "delegated" by writing @bors delegate+ or @bors delegate=username. This will allow the PR author to approve the PR by issuing @bors commands like the ones above (but this privilege is limited to the single PR).

Reverts

If a merged PR is found to have caused a meaningful unanticipated regression, the best policy is to revert it quickly and re-land it later once a fix and regression test are added.

A "meaningful regression" in this case is up to the judgment of the person approving the revert. As a rule of thumb, this would be a bug in a stable or otherwise important feature that causes code to stop compiling, changes runtime behavior, or triggers a (warn-by-default or higher) lint incorrectly in real-world code.

When these criteria are in doubt, and especially if real-world code is affected, revert the PR. This allows bleeding edge users to continue to use and report bugs on HEAD with a higher degree of certainty about where new bugs are introduced.

Before being reverted, a PR should be shown to cause a regression with a fairly high degree of certainty (e.g. bisection on commits, or bisection on nightlies with one or more compiler team members pointing to this PR, or it's simply obvious to everyone involved). Only revert with lower certainty if the issue is particularly critical or urgent to fix.

Creating reverts

The easiest method for creating a revert is to use the "Revert" button on Github. This appears next to the "bors merged commit abcd" message on a pull request, and creates a new pull request.

Alternatively, a revert commit can be created using the git CLI and then uploaded as a pull request:

$ git revert -m 1 62d5bee

It's polite to tag the author and reviewer of the original PR so they know what's going on. You can use the following message template:

Reverts rust-lang/rust#123456
cc @author @reviewer

This revert is based on the following report of a regression caused by this PR:
<link to issue or comment(s)>

In accordance with the compiler team [revert policy], PRs that cause meaningful
regressions should be reverted and re-landed once the regression has been fixed
(and a regression test has been added, where appropriate).
[revert policy]: https://forge.rust-lang.org/compiler/reviews.html#reverts

Fear not! Regressions happen. Please rest assured that this does not
represent a negative judgment of your contribution or ability to contribute
positively to Rust in the future. We simply want to prioritize keeping existing
use cases working, and keep the compiler more stable for everyone.

r? compiler

If you have r+ privileges, you can self-approve a revert.

Generally speaking, reverts should have elevated priority and match the rollup status of the PR they are reverting. If a non-rollup PR is shown to have no impact on performance, it can be marked rollup=always.

Forward fixes

Often it is tempting to address a regression by posting a follow-up PR that, rather than reverting the regressing PR, instead augments the original in small ways without reverting its changes overall. However, if real-world users have reported being affected, this practice is strongly discouraged unless one of the following is true:

• A high-confidence fix is already in the bors queue.
• The regression has made it to a release branch (beta or stable) and a backport is needed. Often the "smallest possible change" is desired for a backport. The offending PR may or may not still be reverted on the main branch; this is left to the discretion of someone who can r+ it.

While it can feel like a significant step backward to have your PR reverted, in most cases it is much easier to land the PR a second time once a fix can be confirmed. Allowing a revert to land takes pressure off of you and your reviewers to act quickly and gives you time to address the issue fully.

Rollups

All reviewers are strongly encouraged to explicitly mark a PR as to whether or not it should be part of a rollup with one of the following:

• rollup=always: These PRs are very unlikely to break tests or have performance implications. Example scenarios:
  • Changes are limited to documentation, comments, etc. that is highly unlikely to fail a build.
  • Changes cannot have performance implications.
  • Your PR is not landing possibly-breaking or behavior altering changes.
    • Feature stabilization without other changes is likely fine to rollup, though.
• rollup=maybe: This is the default if you do not specify a rollup status. Use this if you don't have much confidence that it won't break tests. This can be used if you aren't sure if it should be one of the other categories. Since this is the default, there is usually no need to explicitly specify this, unless you are un-marking the rollup level from a previous command.
• rollup=iffy: Use this for mildly risky PRs (more risky than "maybe"). Example scenarios:
  • The PR is large and non-additive (note: adding 2000 lines of completely new tests is fine to rollup).
  • Messes too much with:
    • LLVM or code generation
    • bootstrap or the build system
    • build-manifest
  • Has platform-specific changes that are not checked by the normal PR checks.
  • May be affected by MIR migrate mode.
• rollup=never: This should never be included in a rollup (please include a comment explaining why you have chosen this). Example scenarios:
  • May have performance implications.
  • May cause unclear regressions (we would likely want to bisect to this PR specifically, as it would be hard to identify as the cause from a rollup).
  • Has a high chance of failure.
  • Is otherwise dangerous to rollup.

Note:
@bors rollup is equivalent to @bors rollup=always
@bors rollup- is equivalent to @bors rollup=never

Priority

Reviewers are encouraged to set one of the rollup statuses listed above instead of setting priority. Bors automatically sorts based on the rollup status (never is the highest priority, always is the lowest), and also by PR age. If you do change the priority, please use your best judgment to balance fairness with other PRs.

The following is some guidance for setting priorities:

• 1-5
  • P-high issue fixes
  • Toolstate fixes
  • Reverts containing the above
  • Beta-nominated PRs
  • Submodule/Subtree updates
• 5+
  • P-critical issue fixes
• 10+
  • Bitrot-prone PRs (particularly very large ones that touch many files)
  • Urgent PRs
  • Beta backports
• 20+
  • High priority that needs to jump ahead of any rollups
  • Fixes or changes something that has a high risk of being re-broken by another PR in the queue.
• 1000
  • Absolutely critical fixes
  • Release promotions

Expectations for r+

bors privileges are binary: the bot doesn't know which code you are familiar with and what code you are not. They must therefore be used with discretion. Do not r+ code that you do not know well -- you can definitely review such code, but try to hand off reviewing to someone else for the final r+.

Similarly, never issue a r=username command unless that person has done the review, and the code has not changed substantially since the review was done. Rebasing is fine, but changes in functionality typically require re-review (though it's a good idea to try and highlight what has changed, to help the reviewer).

So you want to add a new (stable) option to rustc

So you want to add a new command-line flag to rustc. What is the procedure?

Is this a perma-unstable option?

The first question to ask yourself is:

• Is this a "perma-unstable" option meant only for debugging rustc (e.g., -Ztreat-err-as-bug)?

If so, you can just add it in a PR, no check-off is required beyond ordinary review.

Other options

If this option is meant to be used by end-users or to be exposed on the stable channel, however, it represents a "public commitment" on the part of rustc that we will have to maintain, and hence there are a few more details to take care of.

There are two main things to take care of, and they can proceed in either order, but both must be completed:

• Proposal and check-off
• Implementation and documentation

Finally, some options begin as unstable and only get stabilized over time, in which case you will also need:

• Tracking issue and stabilization

Proposal and check-off

The "proposal" part describes the motivation and design of the new option you wish to add. It doesn't necessarily have to be very long. It takes the form of a Major Change Proposal.

The proposal should include the following:

• Motivation: what is this flag used for?
• Design: What input does the flag take and what is its observable effect?
• Implementation notes: You don't have to talk about the implementation normally, but if there are any key things to note (i.e., it was very invasive to implement), you night note them here.
• Precedent, links, and related material: Are similar flags available on other compilers/linkers/tools, like clang or lld?
• Alternatives, concerns, and key decisions: Were there any alernatives considered? If so, why did you pick this design?

Note that it is fine if you don't have any implementation notes, precedent, or alternatives to discuss.

Also, one good approach to writing the MCP is basically to write the documentation you will have to write anyway to explain to users how the option works, and then add any additional notes on alternatives and so forth that are required.

Once you've written up the proposal, you can open a MCP issue. But note that since this MCP is promoting a permanent change, a full compiler-team FCP is required, and not just a "second". This can be done by @rfcbot fcp merge by a team member.

Implementation, documentation

Naturally your new option will also have to be implemented. You can implement the option and open up a PR. Often, this implementation work actually happens before the MCP is created, and that's fine -- we'll just ask you to open an MCP with the write-up.

See the Command-line Arguments chapter in the rustc dev guide for guidelines on how to name and define a new argument.

A few notes that are sometimes overlooked:

• Many options begin as "unstable" options, either because they use -Z or because they require -Zunstable-options to use.
• You should document the option. Often this documentation can just be copied from the MCP text. Where you add this documentation depends on whether the option is available on stable Rust:
  • If it is unstable, then document the option in the Unstable Book, whose sources are in src/doc/unstable-book.
  • Once the option is stabilized, it should be documented in the Rustc book, whose sources as in src/doc/rustc.

Stabilization and tracking issue

Typically options begin as unstable, meaning that they are either used with -Z or require -Zunstable-options.

Once the issue lands we should create a tracking issue that links to the MCP and where stabilization can be proposed.

Stabilization generally proceeds when the option has a seen a bit of use and the implementation seems to be working as expected for its intended purpose.

Remember that when stabilization occurs, documentation should be moved from the Unstable Book to the Rustc Book.

Major Change Proposals

Introduced in RFC 2904, a "major change proposal" is a lightweight form of RFC that the compiler team uses for architectural changes that are not end-user facing. (It can also be used for small user-facing changes like adding new compiler flags, though in that case we also require an rfcbot fcp to get full approval from the team.) Larger changes or modifications to the Rust language itself require a full RFC (the latter fall under the lang team's purview).

Motivation

As the compiler grows in complexity, it becomes harder and harder to track what's going on. We don't currently have a clear channel for people to signal their intention to make "major changes" that may impact other developers in a lightweight way (and potentially receive feedback).

Our goal is to create a channel for signaling intentions that lies somewhere between opening a PR (and perhaps cc'ing others on that PR) and creating a compiler team design meeting proposal or RFC.

Goals

Our goals with the MCP are as follows:

• Encourage people making a major change to write at least a few paragraphs about what they plan to do.
• Ensure that folks in the compiler team are aware the change is happening and given a chance to respond.
• Ensure that every proposal has a "second", meaning some expert from the team who thinks it's a good idea.
• Ensure that major changes have an assigned and willing reviewer.
• Avoid the phenomenon of large, sweeping PRs landing "out of nowhere" onto someone's review queue.
• Avoid the phenomenon of PRs living in limbo because it's not clear what level of approval is required for them to land.

Major Change Proposals

If you would like to make a major change to the compiler, the process is as follows:

• Open a tracking issue on the rust-lang/compiler-team repo using the major change template.
  • A Zulip topic in the stream #t-compiler/major changes will automatically be created for you by a bot.
  • If concerns are raised, you may want to modify the proposal to address those concerns.
  • Alternatively, you can submit a design meeting proposal to have a longer, focused discussion.
• To be accepted, a major change proposal needs three things:
  • One or more reviewers, who commit to reviewing the work. This can be the person making the proposal, if they intend to mentor others.
  • A second, a member of the compiler team or a contributor who approves of the idea, but is not the one originating the proposal.
  • A final comment period (a 10 day wait to give people time to comment).
    • The FCP can be skipped if the change is easily reversed and/or further objections are considered unlikely. This often happens if there has been a lot of prior discussion, for example.
• Once the FCP completes, if there are no outstanding concerns, PRs can start to land.
  • If those PRs make outward-facing changes that affect stable code, then either the MCP or the PR(s) must be approved with a rfcbot fcp merge comment.

Conditional acceptance

Some major change proposals will be conditionally accepted. This indicates that we'd like to see the work land, but we'd like to re-evaluate the decision of whether to commit to the design after we've had time to gain experience. We should try to be clear about the things we'd like to evaluate, and ideally a timeline.

Deferred or not accepted

Some proposals will not be accepted. Some of the possible reasons:

• You may be asked to do some prototyping or experimentation before a final decision is reached
• The idea might be reasonable, but there may not be bandwidth to do the reviewing, or there may just be too many other things going on.
• The idea may be good, but it may be judged that the resulting code would be too complex to maintain, and not worth the benefits.
• There may be flaws in the idea or it may not sufficient benefit.

What happens if someone opens a PR that seems like a major change without doing this process?

The PR should be closed or marked as blocked, with a request to create a major change proposal first.

If the PR description already contains suitable text that could serve as an MCP, then simply copy and paste that into an MCP issue. Using an issue consistently helps to ensure that the tooling and process works smoothly.

Can I work on code experimentally before a MCP is accepted?

Of course! You are free to work on PRs or write code. But those PRs should be marked as experimental and they should not land, nor should anyone be expected to review them (unless folks want to).

What constitutes a major change?

The rough intuition is "something that would require updates to the rustc-dev-guide or the rustc book". In other words:

• Something that alters the architecture of some part(s) of the compiler, since this is what the rustc-dev-guide aims to document.
• A simple change that affects a lot of people, such as altering the names of very common types or changing coding conventions.
• Adding a compiler flag or other public facing changes, which should be documented (ultimately) in the rustc book. This is only appropriate for "minor" tweaks, however, and not major things that may impact a lot of users. (Also, public facing changes will require a full FCP before landing on stable, but an MCP can be a good way to propose the idea.)

Note that, in some cases, the change may be deemed too big and a full FCP or RFC may be required to move forward. This could occur with significant public facing change or with sufficiently large changes to the architecture. The compiler team leads can make this call.

Note that whether something is a major change proposal is not necessarily related to the number of lines of code that are affected. Renaming a method can affect a large number of lines, and even require edits to the rustc-dev-guide, but it may not be a major change. At the same time, changing names that are very broadly used could constitute a major change (for example, renaming from the tcx context in the compiler to something else would be a major change).

Public-facing changes require rfcbot fcp

The MCP "seconding" process is only meant to be used to get agreement on the technical architecture we plan to use. It is not sufficient to stabilize new features or make public-facing changes like adding a -C flag. For that, an rfcbot fcp is required (or perhaps an RFC, if the change is large enough).

For landing compiler flags in particular, a good approach is to start with an MCP introducing a -Z flag and then "stabilize" the flag by moving it to -C in a PR later (which would require rfcbot fcp).

Major change proposals are not sufficient for language changes or changes that affect cargo.

Steps to open a MCP

• Open a tracking issue on the rust-lang/compiler-team repo using the major change template.
• Create a Zulip topic in the stream #t-compiler/major changes:
  • The topic should be named something like "modify the whiz-bang component compiler-team#123", which describes the change and links to the tracking issue.
  • The stream will be used for people to ask questions or propose changes.

What kinds of comments should go on the tracking issue in compiler-team repo?

Please direct technical conversation to the Zulip stream.

The compiler-team repo issues are intended to be low traffic and used for procedural purposes. Note that to "second" a design or offer to review, you should be someone who is familiar with the code, typically but not necessarily a compiler team member or contributor.

• Announcing that you "second" or approve of the design.
• Announcing that you would be able to review or mentor the work.
• Noting a concern that you don't want to be overlooked.
• Announcing that the proposal will be entering FCP or is accepted.

How does one register as reviewer, register approval, or raise an objection?

These types of procedural comments can be left on the issue (it's also good to leave a message in Zulip). See the previous section.

Who decides whether a concern is unresolved?

Usually the experts in the given area will reach a consensus here. But if there is some need for a "tie breaker" vote or judgment call, the compiler-team leads make the final call.

What are some examples of major changes from the past?

Here are some examples of changes that were made in the past that would warrant the major change process:

• overhauling the way we encode crate metadata
• merging the gcx, tcx arenas
• renaming a widely used, core abstraction, such as the Ty type
• introducing cargo pipelining
• adding a new -C flag that exposes some minor variant

What are some examples of things that are too big for the major change process?

Here are some examples of changes that are too big for the major change process, or which at least would require auxiliary design meetings or a more fleshed out design before they can proceed:

• introducing incremental or the query system
• introducing MIR or some new IR
• introducing parallel execution
• adding ThinLTO support

What are some examples of things that are too small for the major change process?

Here are some examples of things that don't merit any MCP:

• adding new information into metadata
• fixing an ICE or tweaking diagnostics
• renaming "less widely used" methods

When should Major Change Proposals be closed?

Major Change Proposals can be closed:

• by the author, if they have lost interest in pursuing it.
• by a team lead or expert, if there are strong objections from key members of the team that don't look likely to be overcome.
• by folks doing triage, if there have been three months of inactivity. In this case, people should feel free to re-open the issue if they would like to "rejuvenate" it.

Membership

This team discusses membership in the compiler team. There are currently two levels of membership:

• contributors: regular contributors with r+ rights, bot privileges, and access to infrastructure
• full members: full members who vote on RFCs

The path to membership

People who are looking to contribute to the compiler typically start in one of two ways. They may tackle "one off" issues, or they may get involved in some kind of existing working group. They don't know much about the compiler yet and have no particular privileges. They are assigned to issues using the triagebot and (typically) work with a mentor or mentoring instructions.

Compiler team contributors

Once a working group participant has been contributing regularly for some time, they can be promoted to the level of a compiler team contributor (see the section on how decisions are made below). This title indicates that they are someone who contributes regularly.

It is hard to define the precise conditions when such a promotion is appropriate. Being promoted to contributor is not just a function of checking various boxes. But the general sense is that someone is ready when they have demonstrated three things:

• "Staying power" -- the person should be contributing on a regular basis in some way. This might for example mean that they have completed a few projects.
• "Independence and familiarity" -- they should be acting somewhat independently when taking on tasks, at least within the scope of the working group. They should plausibly be able to mentor others on simple PRs.
• "Cordiality" -- contributors will be members of the organization and are held to a higher standard with respect to the Code of Conduct. They should not only obey the letter of the CoC but also its spirit.

Being promoted to contributor implies a number of privileges:

• Contributors have r+ privileges and can do reviews (they are expected to use those powers appropriately, as discussed previously). They also have access to control perf/rustc-timer and other similar bots.
• Contributors are members of the organization so they can modify labels and be assigned to issues.
• Contributors are a member of the rust-lang/compiler team on GitHub, so that they receive pings when people are looking to address the team as a whole.
• Contributors are listed on the rust-lang.org web page.

It also implies some obligations (in some cases, optional obligations):

• Contributors will be asked if they wish to be added to the reviewer rotation.
• Contributors are held to a higher standard than ordinary folk when it comes to the Code of Conduct.

Full members

As a contributor gains in experience, they may be asked to become a compiler team member. This implies that they are not only a regular contributor, but are actively helping to shape the direction of the team or some part of the compiler (or multiple parts).

• Compiler team members are the ones who select when people should be promoted to compiler team contributor or to the level of member.
• Compiler team members are consulted on FCP decisions (which, in the compiler team, are relatively rare).
• There will be a distinct GitHub team containing only the compiler team members, but the name of this team is "to be determined".
• Working groups must always include at least one compiler team member as a lead (though groups may have other leads who are not yet full members).

How promotion decisions are made

Promotion decisions (from participant to contributor, and from contributor to member) are made by having an active team member send an e-mail to the alias compiler-private@rust-lang.org. This e-mail should include:

• the name of the person to be promoted
• a draft of the public announcement that will be made

Compiler-team members should send e-mail giving their explicit assent, or with objections. Objections should always be resolved before the decision is made final. E-mails can also include edits or additions for the public announcement.

To make the final decision:

• All objections must be resolved.
• There should be a "sufficient number" (see below) of explicit e-mails in favor of addition (including the team lead).
• The nominator (or some member of the team) should reach out to the person in question and check that they wish to join.

We do not require all team members to send e-mail, as historically these decisions are not particularly controversial. For promotion to a contributor, the only requirement is that the compiler team lead agrees. For promotion to a full member, more explicit mails in favor are recommended.

Once we have decided to promote, then the announcement can be posted to internals, and the person added to the team repository.

Not just code

It is worth emphasizing that becoming a contributor or member of the compiler team does not necessarily imply writing PRs. There are a wide variety of tasks that need to be done to support the compiler and which should make one eligible for membership. Such tasks would include organizing meetings, participating in meetings, bisecting and triaging issues, writing documentation, working on the rustc-dev-guide. The most important criteria for elevation to contributor, in particular, is regular and consistent participation. The most important criteria for elevation to member is actively shaping the direction of the team or compiler.

Alumni status

If at any time a current contributor or member wishes to take a break from participating, they can opt to put themselves into alumni status. When in alumni status, they will be removed from Github aliases and the like, so that they need not be bothered with pings and messages. They will also not have r+ privileges. Alumni members will however still remain members of the GitHub org overall.

People in alumni status can ask to return to "active" status at any time. This request would ordinarily be granted automatically barring extraordinary circumstances.

People in alumni status are still members of the team at the level they previously attained and they may publicly indicate that, though they should indicate the time period for which they were active as well.

Changing back to contributor

If desired, a team member may also ask to move back to contributor status. This would indicate a continued desire to be involved in rustc, but that they do not wish to be involved in some of the weightier decisions, such as who to add to the team. Like full alumni, people who were once full team members but who went back to contributor status may ask to return to full team member status. This request would ordinarily be granted automatically barring extraordinary circumstances.

Automatic alumni status after 6 months of inactivity

If a contributor or a member has been inactive in the compiler for 6 months, then we will ask them if they would like to go to alumni status. If they respond yes or do not respond, they can be placed on alumni status. If they would prefer to remain active, that is also fine, but they will get asked again periodically if they continue to be inactive.

Prioritization

This section documents the processes of the prioritization WG.

Prioritization WG - Procedure

This document details the procedure the WG-prioritization follows to fill the agenda for the weekly meeting of T-compiler. The working group focuses mainly on triaging T-compiler regressions, identifying possibly critical (and thus potential release blocker) issues and building the agenda for the weekly T-compiler meeting summarizing the main points to be discussed.

General issues review process

• Check the status of the issue
• Try moving it forward if possible (ex. stimulate further comments from the issue author / reviewer)
• Ask for more info if it's needed
• Is there an MCVE for the issue already?
• Check if it's a regression and label it accordingly (regression-* labels)
• Figure out the area the issue belongs and label it accordingly (A-* labels)
• Ping notify groups or relevant teams
• Assign if possible
• Nominate the issue if it's unclear and needs to be discussed

Generating the T-compiler meeting's agenda

The T-compiler agenda is generated from a template (available on HackMD or Github). We suggest working the following steps in this order:

Prepare agenda content

1. Add T-compiler labels where appropriate

• Issues labeled with I-prioritize
• Pull requests nominated for the stable release channel backport
• Pull requests nominated for the beta release channel backport
• Issues labeled I-compiler-nominated (i.e. needing a T-compiler discussion)
• Pull requests waiting on a team's feedback
• Issues classified with priority P-high

2. Assign a priority label to issues where needed

Regressions labeled with I-prioritize are signaling that a priority assessment is waiting. When this label is added to an issue, the triagebot creates automatically a notification for @WG-prioritization members on the Zulip stream.

To assign a priority, we replace the I-prioritize label with one of P-critical, P-high, P-medium or P-low and adding a succinct comment to link the Zulip discussion where the issue prioritization occurred, example of a template for the comment:

WG-prioritization assigning priority (Zulip discussion).

@rustbot label -I-prioritize +P-XXX

Ideally, we want all T-compiler issues with a I-prioritize label to have a priority assigned, or strive to reach this goal: sometimes different factors are blocking issues from being assigned a priority label, either because the report or the context is unclear or because cannot be reproduced and an MCVE would help. Don't hesitate to ask for clarifications to the issue reporter or ping the ICEbreaker team when an ICE ("Internal Compiler Errors") needs a reduction (add a comment on the issue with @rustbot ping icebreakers-cleanup-crew)

Keep an eye also on regressions (stable, beta and nightly), ideally they should an assignee.

3. Accept MCPs

An MCP is a Major Change Proposal, in other words a change to the rust compiler that needs a bit more thought and discussion within the compiler team than a pull request. The life cycle of an MCP is described in the documentation. The relevant part for the WG-Prioritization is keeping an eye on them and accept all MCPs that have been on final-comment-period for 10 or more days.

To accept an MCP, remove final-comment-period label, add major-change-accepted label and close the issue. A notification to the relevant Zulip topic (in this stream) will be automatically sent by the triagebot.

Generate the meeting's agenda

Run triagebot's CLI to generate the agenda. You need to clone https://github.com/rust-lang/triagebot (there is no official prepackaged release for this tool) and export two environment variables: GITHUB_API_TOKEN and optionally a GOOGLE_API_KEY to access a public Google calendar (if this env var is not found, meetings should be manually copy&pasted from here).

To generate the meeting's agenda, run:

$ cargo run --bin prioritization-agenda

Copy the content of the generated agenda on HackMD. This will be our starting point.

Add performance logs

Paste the markdown file of this week performance triage logs to the agenda and clean it up a little bit removing emojis (to make the text readable when pasted on Zulip).

Announce the meeting on Zulip

About two hours before the scheduled meeting, create a new topic on the Zulip stream #t-compiler/meetings titled "[weekly] YYYY-MM-DD" using the the following message template:

Hi @*T-compiler/meeting*; the triage meeting will happen tomorrow in about 2 hours.
*WG-prioritization* has done pre-triage in #**t-compiler/wg-prioritization/alerts**
@*WG-prioritization* has prepared the [meeting agenda](link_to_hackmd_agenda)

Working group checkins for today:
- @**WG-foo** by @**person1**
- @**WG-bar** by @**person2**

Working Group checkins rotation are generated by a script at this page (TODO: script is outdated and could probably be merged into the triagebot CLI code).

Checkins about the progress of working groups are not mandatory but we rotate them all to be sure we don't miss on important progresses.

Add details to the Agenda

1. Summarize stable/beta nominations

These are pull requests that the compiler team might want to backport to a release channel. Example a stable-to-beta-regression fix might want to be backported to the beta release channel. A stable-to-stable-regression fix particularly annoying might warrant a point release (i.e. release a 1.67.1 after a 1.67.0).

Follow the General issues review process.

2. Summarize PRs waiting on team

These are pull requests waiting on a discussion / decision from T-compiler (sometimes more than one team).

Try to follow the General issues review process. Explicitly nominate any issue that can be quickly resolved in a triage meeting.

3. Fill up the "Oldest PRs waiting for review"

This is probably the less automatable part of the agenda (and likely the least fun). The triagebot will emit a list of 50 pull requests ordering them by least recent update. The idea is to issue mentions to assigned reviewers during the meeting ensuring that they stay on top of them. We usually try to keep the number of these mentions to around 5 for each meeting.

There are two human factors here to keep in consideration:

• Pull requests reviewers are volunteers, we respect and appreciate their work. We don't want to remind them too often that there is a pile of pull requests waiting on them. Therefore we usually wait 2 or 3 weeks before reminding them about that pull requests. It seems like a long time to wait but let's not forget what contributors accomplish in the meanwhile! Anyway, we are trying to find ways to improve on these metrics.
• Contributors taking their time to submit a pull request deserve equally our appreciation so we try to not have them wait too long for a review or they will lose context about their work (or motivation to drive the contribution to completion).

Striking a balance between these two diverging forces requires some empathy and "tribal knowledge" that comes with practice. Other factors can be blocking a pull request progress:

• The review is shared with another team (i.e. Team 1 says "OK", now waiting on Team 2)
• The alternating labels S-waiting-on-review and S-waiting-on author handling the life cycle of a pull request are not promptly applied. A pull request that is ready to be reviewed but it's not labeled S-waiting-on-review is idling for no purpose.

4. Add some context to P-critical and P-high regressions without an assignee

Try to follow the General issues review process.

5. Summarize I-compiler-nominated issues

Issues labeled with I-compiler-nominated generally are nominated to specifically have the compiler team dedicate them a special slice of the meeting (generally towards the end). After the discussion, add a comment on Github linking the Zulip message where the discussion started (so everyone can read). T-compiler sometimes writes a summary of the discussion on the issue itself.

Try to follow the General issues review process:

• Check if an issue needs a discussion and add the label I-compiler-nominated
• When added to the agenda, add some context:
  • Who the assignee is
  • Is this an issue or a pull request: if it's an issue, does it have a pull request that fixes it?
  • Why was it nominated
  • Other important details

6. Final review before the meeting

Re-run the triagebot CLI script and update the agenda on HackMD with new data (if any). This is useful when there are last second changes affecting the agenda content.

Follow-ups after meeting

The meeting is over! Time to cleanup a little bit.

• Lock the agenda file on HackMD assigning write permissions to Owners. Download the markdown file and commit it to this repository.

• Remove the to-announce label from MCPs, unless this label was added exactly during the meeting (and therefore will be seen during the following meeting).

• Remove to-announce FCPs from rust repo, compiler-team repo and forge repo, same disclaimer as before.

• Accept or decline beta nominated and stable nominated backports that have been accepted during the meeting. For more info check T-release backporting docs

  • To accept a backport, add a {beta,stable}-accepted label and keep the {beta,stable}-nominated label. Other automated procedures will process these pull requests, it's important to leave both labels. Add a comment on Github linking the Zulip discussion.
  • To decline a backport, simply remove {beta,stable}-nominated label. Add a comment on Github explaining why the backport was declined and link the Zulip discussion.

• Remove I-compiler-nominated label from issues that were discussed. Sometimes not all nominated issues are discussed (because of time constraints). In this case the I-compiler-nominated will stick until next meeting.

• Create a new agenda stub for the following week using our template and post the link on Zulip, so it's available for people if they want to add content during the week.

Priority levels

As the compiler team's resources are limited, the prioritization working group's main goal is to identify the most relevant issues to work on, so that the compiler team can focus on what matters the most.

Words used in this document:

issue refers to bugs and feature requests that are nominated for prioritization, by flagging the I-prioritize label as described below.

This document will define what each label means, and what strategy for each label will be used.

Labels

Labeling an issue as I-prioritize starts the prioritization process, which will end by removing the I-prioritize label and appending one of the 4 labels we will discuss below:

• P-critical
• P-high
• P-medium
• P-low

Each of these labels defines a strategy the team will adopt regarding:

• The amount of focus a given issue will receive
• How members of the community can get involved

P-critical

A P-critical issue is a potentially blocker issue.

The Working Group will keep track of these issues and will remind the compiler team on a weekly basis during the triage meeting.

Examples of things we typically judge to be “critical” bugs:

• Regressions where code that used to compile no longer does
  • Mitigating conditions that may lower priority:
    • If the code should never have compiled in the first place (but if the regression affects a large number of crates, this may indicate that we need a warning period)
    • If the code in question is theoretical and considered unlikely to exist in the wild, or if it only exists in small, unmaintained packages that are not widely used
  • If a regression has been in stable for a release or two (either because we are still awaiting a fix, or because the bug had laid dormant i.e. undetected), we typically lower the priority as well, because by that time, if the users have not raised a ruckus about the regression, that is a sign that it is inherently not a critical issue. Eg: an issue that would have been P-critical but ended up being P-high
• Regressions where code still compiles but does something different than it used to do (dynamic semantics have changed)
  • Mitigating conditions that may lower priority:
    • If code uses feature that is explicitly not specified (e.g. std::vec::Vec docs state order in which it drops its elements is subject to change)
• Feature-gated features accessible without a feature gate
  • Mitigating conditions that may lower priority:
    • If the pattern is VERY unlikely
• Soundness holes with real-world implications
  • Mitigating conditions that may lower priority:
    • Soundness holes that are difficult to trigger
    • Soundness holes that will not affect stable, e.g. if the hole makes use of a gated unstable feature.
• Diagnostic regressions where the diagnostic is very common and the situation very confusing
• ICEs for common scenarios or code patterns
  • Mitigating conditions that may lower priority:
    • If the code that triggers the ICE also triggers compilation errors, and those errors are emitted before the ICE
    • If the code in question makes use of unstable features, particularly if the ICE requires a feature gate

A P-critical issue will receive the most attention. It must be assigned one or several people as soon as possible, and the rest of the team should do their best to help them out if/when applicable.

P-high

P-high issues are issues that need attention from the compiler team, but not to the point that they need to be discussed at every meeting. They can be P-critical issues that have a mitigating condition as defined above, or important issues that aren't deemed blockers.

Because there are too many P-high issues to fit in every compiler meeting, they should rather be handled asynchronously by the Prioritization WG, in order to help them move forward. They can still occasionally be brought up at meetings when it is deemed necessary.

The effectiveness of the Prioritization WG will be a direct consequence of our ability to draw the line between P-critical and P-high issues. There shouldn't be too many P-critical issues that compiler meetings become unmanageable, but critical issues shouldn't get lost in the list of P-high issues.

P-high issues are issues the teams will mostly work on. We want to make sure they're assigned, and keep an eye on them.

P-medium and P-low

P-medium refer to issues that aren't a priority for the team, and that will be resolved in the long run. Eg issues that will be fixed after a specific feature has landed. They are issues we would mentor someone interested in fixing. They will remain in this state until someone complains, a community member fixes it, or it gets fixed by accident.

P-low refer to issues issue that the compiler team doesn't plan to resolve, but are still worth fixing.

Notification groups

The compiler team has a number of notification groups that we use to ping people and draw their attention to issues. Notification groups are setup so that anyone can join them if they want.

Creating a notification group

If you'd like to create a notification group, here are the steps. First, you want to get approval from the compiler team:

• Propose the group by preparing a Major Change Proposal. If your group is not analogous to some existing group, it is probably a good idea to ping compiler team leads before-hand or as part of the MCP.
• The MCP should specify what GitHub label will be associated with the notification group. Often this is an existing label, such as O-Windows.

Once the MCP is accepted, here are the steps to actually create the group. In some cases we include an example PR from some other group.

• File a tracking issue in the rust-lang/compiler-team repository to collect your progress.
• Create a PR against the rust-lang/team repository adding the notification group. Example PR.
• Configure the rust-lang/rust repository to accept triagebot commands for this group. Example PR.
• Create a PR for the rustc-dev-guide amending the notification group section to mention your group.
• Create a sample PR for the rust-lang/team repository showing how one can add oneself. This will be referenced by your blog post to show people how to join. Example PR.
• Create a Zulip stream for the notification group. If you don't have the permission to do, you can ask on #t-compiler/wg-meta.
• Write an announcement blog post for Inside Rust and open a PR against blog.rust-lang.org. Example PR.

Compiler-team Triage Meeting

What is it?

The triage meeting is a weekly meeting where we go over the open issues, look at regressions, consider beta backports, and other such business. In the tail end of the meeting, we also do brief check-ins with active working groups to get an idea what they've been working on.

When and where is it?

See the compiler team meeting calendar for the canonical date and time. The meetings take place in the #t-compiler stream on the rust-lang Zulip.

Where can I lean more?

The meeting procedure is documented in rust-lang/rust#54818.

The working group check-in schedule is available on the compiler-team website.

Compiler-team Steering Meeting

What is it?

The "steering meeting" is a weekly meeting dedicated to planning and high-level discussion. The meeting operates on a repeating schedule:

• Week 1: Planning
• Week 2: Technical or non-technical discussion
• Week 3: Technical or non-technical discussion
• Week 4: Non-technical discussion

The first meeting of the 4-week cycle is used for planning. The primary purpose of this meeting is to select the topics for the next three meetings. The topics are selected from a set of topic proposals, which must be uploaded and available for perusal before the meeting starts. The planning meeting is also an opportunity to check on the "overall balance" of our priorities.

The remaining meetings are used for design or general discussion. Weeks 2 and 3 can be used for technical or non-technical discussion; it is also possible to use both weeks to discuss the same topic, if that topic is complex. Week 4 is reserved for non-technical topics, so as to ensure that we are keeping an eye on the overall health and functioning of the team.

Where do proposals come from?

The team accepts proposals via an open submission process, which is documented on its own page

Announcing the schedule

After each planning meeting, the topics for the next three weeks are added to the compiler-team meeting calendar and a blog post is posted to the Inside Rust blog.

When and where is it?

See the compiler team meeting calendar for the canonical date and time. The meetings take place in the #t-compiler stream on the rust-lang Zulip.

Submitting a proposal

If you would like to submit a proposal to the steering meeting for group discussion, read on! This page has all the details.

TL;DR

In short, all you have to do is

• open an issue on the compiler-team repository
  • use the template for meeting proposals
  • you only need a few sentences to start, but by the time the meeting takes place we typically expect a more detailed writeup, e.g. using this template

You don't have to have a lot of details to start: just a few sentences is enough. But, especially for technical design discussions, we will typically expect that some form of more detailed overview be made available by the time the meeting takes place.

Examples of good candidates for discussing at the steering meeting

Here are some examples of possible technical topics that would be suitable for the steering meeting:

• A working group has an idea to refactor the HIR to make some part of their job easier. They have sketched out a proposal and would like feedback.
• Someone has encountered a problem that is really hard to solve with the existing data structures. They would like feedback on a good solution to their problem.
• Someone has done major refactoring work on a PR and they would like to be able to explain the work they did and request review.

Steering meetings are also a good place to discuss other kinds of proposals:

• A proposal to move some part of the compiler into an out-of-tree crate.
• A proposal to start a new working group.

Note that a steering meeting is not required to create a new working group or an out-of-tree crate, but it can be useful if the proposal is complex or controversial, and you would like a dedicated time to talk out the plans in more detail.

Criteria for selection

When deciding the topics for upcoming meetings, we must balance a number of things:

• We don't want to spend time on design work unless there are known people who will implement it and support it; this includes not only the "main coder" but also a suitable reviewer.
• We don't want to take on "too many" tasks at once, even if there are people to implement them.
• We also don't want to have active projects that will be "stepping on each others' toes", changing the same set of code in deep ways.

Meetings are not mandatory

It is perfectly acceptable to choose not to schedule a particular slot. This could happen if (e.g.) there are no proposals available or if nothing seems important enough to discuss at this moment. Note that, to keep the "time expectations" under control, we should generally stick to the same 4-week cycle and simply opt to skip meetings, rather than (e.g.) planning things at the last minute.

Adding a proposal

Proposals can be added by opening an issue on the compiler-team repository. There is an issue template for meeting proposals that gives directions. The basic idea is that you open an issue with a few sentences describing what you would like to talk about.

Some details that might be useful to include:

• how complex of a topic you think this is
• people in the compiler team that you think should be present for the meeting

Expectations for the meeting

By the time the meeting takes place, we generally would prefer to have a more detailed write-up or proposal. You can find a template for such a proposal here. This should be created in the form of a hackmd document -- usually we will then update this document with the minutes and consensus from the meeting. The final notes are then stored in the minutes directory of the compiler-team repository.

Expectations for a non-technical proposal

The requirements for non-technical proposals are somewhat looser. A few sentences or paragraphs may well suffice, if it is sufficient to understand the aims of the discussion.

Frequently asked questions

What happens if there are not enough proposals? As noted above, meetings are not mandatory. If there aren't enough proposals in some particular iteration, then we can just opt to not discuss anything.

How to run the planning meeting

Week of the meeting

• Announce the meeting in the triage meeting
• Skim over the list of proposals and ping people who have open proposals to get their availability over the next few weeks

Day of the meeting

• Create a design meeting YYYY.MM.DD topic
  • Ping @t-compiler/meeting, ideally 1h or so before the meeting actually starts, to remind people
• At the time of the meeting, return to the topic
  • Ping @t-compiler/meeting to let people know the meeting is starting
• We typically begin with a 5min announcement period
• Visit the compiler-team repository to get a list of proposed meetings

To actually make the final selection, we recommend

• First, try to identify topics that are clear non-candidates
  • for example, sometimes more investigative work (e.g., data gathering) is needed
  • try to identify people to do those tasks
  • other issues may be out of date, or clear non-starters, and they can be closed
• Next tackle technical design meetings, then non-technical
  • Typical ratio is 2 technical, 1 non-technical, but this is not set in stone
  • It's ok to have fewer than 3 meetings

Announce the meetings

For each scheduled meeting, create a calendar event:

• invite key participants to the meeting
• set the location to #t-compiler, Zulip
• include a link to the design meeting issue in the event

In the relevant issues, add the meeting-scheduled label and add a message like:

In today's [planning meeting], we decided to schedule this meeting for **DATE**.

[Calendar event]

[planning meeting]: XXX link to Zulip topic
[Calendar event]: XXX link to calendar event

You can get the link to the calendar event by clicking on the event in google calendar and selecting "publish".

Publish a blog post

Add a blog post to the Inside Rust blog using the template found on the compiler-team repository.

How to run the design meeting

Week of the meeting

• Announce the meeting in the triage meeting
• Skim over the list of proposals and ping people who have open proposals to get their availability over the next few weeks
• Make sure that a write-up is available and nag the meeting person otherwise

Day of the meeting

• Create a design meeting YYYY.MM.DD topic
  • Ping @t-compiler/meeting, ideally 1h or so before the meeting actually starts, to remind people
  • Include a link to the design meeting write-up
• At the time of the meeting, return to the topic
  • Ping @t-compiler/meeting to let people know the meeting is starting
  • Include a link to the design meeting write-up
• We typically begin with a 5min announcement period

To guide the meeting, create a shared hackmd document everyone can view (or adapt an existing one, if there is a write-up). Use this to help structure the meeting, document consensus, and take live notes. Try to ensure that the meeting ends with sort of consensus statement, even if that consensus is just "here are the problems, here is a space of solutions and their pros/cons, but we don't have consensus on which solution to take".

After the meeting

• Post the final contents of the summary hackmd as minutes to the minutes/design-meeting directory in the compiler-team repository
• (Optional) Create an Inside Rust blog post pointing people at the minutes and maybe giving a few notes

crates.io

This section documents the processes of the crates.io team.

Crate removal procedure

If we get a DMCA takedown notice, here's what needs to happen:

Contact Legal

Before removing the crates, get in touch with legal support, currently by emailing the Core team, and ask an opinion from them on the received request and whether we have to comply with it.

Remove relevant version(s) and/or entire crates from crates.io

• Remove it from the database:

  heroku run -a crates-io -- target/release/crates-admin delete-crate [crate-name]
  

  or

  heroku run -a crates-io -- target/release/crates-admin delete-version [crate-name] [version-number]
  

• Remove the crate or version from the index. To remove an entire crate, remove the entire crate file. For a version, remove the line corresponding to the relevant version.

• Remove the crate archive(s) and readme file(s) from S3.

• Invalidate the CloudFront cache:

  aws cloudfront create-invalidation --distribution-id EJED5RT0WA7HA --paths '/*'
  

Remove entire crates from docs.rs

The docs.rs application supports deleting all the documentation ever published of a crate, by running a CLI command. The people who currently have permissions to access the server and run it are:

• docs.rs Team:
  • @pietroalbini
  • @jyn514
• Infrastructure Team:
  • @Mark-Simulacrum
• People with elevated 1password access

You can find the documentation on how to run the command here.

Database maintenance

There are times when Heroku needs to perform a maintenance on our database instances, for example to apply system updates or upgrade to a newer database server.

We must not let Heroku run maintenances during the maintenance window to avoid disrupting production users (move the maintenance window if necessary). This page contains the instructions on how to perform the maintenance with the minimum amount of disruption.

Primary database

Performing maintenance on the primary database requires us to temporarily put the application in read-only mode. Heroku performs maintenances by creating a hidden database follower and switching over to it, so we need to prevent writes on the primary to let the follower catch up.

Maintenance should take less than 5 minutes of read-only time, but we should still announce it ahead of time on our status page. This is a sample message we can use:

The crates.io team will perform a database maintenance on YYYY-MM-DD from hh:mm to hh:mm UTC.

We expect this to take less than 5 minutes to complete. During maintenance crates.io will only be available in read-only mode: downloading crates and visiting the website will still work, but logging in, publishing crates, yanking crates or changing owners will not work.

Primary database checklist

1 hour before the maintenance

1. Go into the Heroku Scheduler and disable the job enqueueing the downloads count updater. You can "disable" it by changing its schedule not to run during the maintenance window. The job uses a lot of database resources, and we should not run it during maintenance.

5 minutes before the maintenance

2. Scale the background worker to 0 instances:

   heroku ps:scale -a crates-io background_worker=0
   

At the start of the maintenance

3. Update the status page with this message:

   Scheduled maintenance on our database is starting.

   We expect this to take less than 5 minutes to complete. During maintenance crates.io will only be available in read-only mode: downloading crates and visiting the website will still work, but logging in, publishing crates, yanking crates or changing owners will not work.

4. Configure the application to be in read-only mode without the follower:

   heroku config:set -a crates-io READ_ONLY_MODE=1 DB_OFFLINE=follower
   

   The follower is removed because while Heroku tries to prevent connections to the primary database from failing during maintenance we observed that the same does not apply to the follower database, and there could be brief periods while the follower is not available.

5. Wait for the application to be redeployed with the new configuration:

   heroku ps:wait -a crates-io
   

6. Run the database maintenance:

   heroku pg:maintenance:run --force -a crates-io
   

7. Wait for the maintenance to finish:

   heroku pg:wait -a crates-io
   

8. Confirm all the databases are online:

   heroku pg:info -a crates-io
   

9. Confirm the primary database fully recovered (should output false):

   echo "SELECT pg_is_in_recovery();" | heroku pg:psql -a crates-io DATABASE
   

10. Switch off read-only mode:

    heroku config:unset -a crates-io READ_ONLY_MODE
    

    WARNING: the Heroku Dashboard's UI is misleading when removing an environment variable. A red badge with a "-" (minus) in it means the variable was successfully removed, it doesn't mean removing the variable failed. Failures are indicated with a red badge with a "x" (cross) in it.

11. Wait for the application to be redeployed with the new configuration:

    heroku ps:wait -a crates-io
    

12. Update the status page and mark the maintenance as completed with this message:

    Scheduled maintenance finished successfully.

    The message is posted right now and not at the end because this is when production users are not impacted by the maintenance anymore.

13. Scale the background worker up again:

    heroku ps:scale -a crates-io background_worker=1
    

14. Confirm the follower database is available:

    echo "SELECT 1;" | heroku pg:psql -a crates-io READ_ONLY_REPLICA
    

15. Enable connections to the follower:

    heroku config:unset -a crates-io DB_OFFLINE
    

16. Re-enable the background job disabled during step 1.

Follower database

Performing maintenance on the follower database doesn’t require any external communication nor putting the application in read-only mode, as we can just redirect all of the follower’s traffic to the primary database. It shouldn’t be done during peak traffic periods though, as we’ll increase the primary database load by doing this.

Follower database checklist

At the start of the maintenance

1. Configure the application to operate without the follower:

   heroku config:set -a crates-io DB_OFFLINE=follower
   

2. Wait for the application to be redeployed with the new configuration:

   heroku ps:wait -a crates-io
   

3. Start the database maintenance:

   heroku pg:maintenance:run --force -a crates-io READ_ONLY_REPLICA
   

4. Wait for the maintenance to finish:

   heroku pg:wait -a crates-io READ_ONLY_REPLICA
   

5. Confirm the follower database is ready:

   heroku pg:info -a crates-io
   

6. Confirm the follower database is responding to queries:

   echo "SELECT 1;" | heroku pg:psql -a crates-io READ_ONLY_REPLICA
   

7. Enable connections to the follower:

   heroku config:unset -a crates-io DB_OFFLINE
   

8. Wait for the application to be redeployed with the new configuration.

   heroku ps:wait -a crates-io
   

docs.rs

docs.rs is a website that hosts documentation for crates published to crates.io.

External Links

• Source code: rust-lang/docs.rs
• Hosted on: docsrs.infra.rust-lang.org (behind the bastion -- how to connect)
• Maintainers: docs.rs team
• Instance metrics (only available to infra team members).
• Application metrics (only available to infra team members).

Add a dependency to the build environment

Rustwide internally uses rustops/crates-build-env as the build environment for the crate. If you want to add a system package for crates to link to, this is place you're looking for.

Preconditions

Docker and docker-compose must be installed. For example, on Debian or Ubuntu:

sudo apt-get install docker.io docker-compose

Getting started

First, clone the crates-build-env and the docs.rs repos:

git clone https://github.com/rust-lang/crates-build-env
git clone https://github.com/rust-lang/docs.rs

Set the path to the directory of your crate. This must be an absolute path, not a relative path! On platforms with coreutils, you can instead use $(realpath ../relative/path) (relative to the docs.rs directory).

YOUR_CRATE=/path/to/your/crate

Add package

Next, add the package to crates-build-env/linux/packages.txt in the correct alphabetical order. This should be the name of a package in the Ubuntu 20.04 Repositories. See the package home page for a full list/search bar, or use apt search locally.

Building the image

Now build the image. This will take a very long time, probably 10-20 minutes.

cd crates-build-env/linux
docker build --tag build-env .

Testing the image

Use the image to build your crate.

cd ../../docs.rs
cp .env.sample .env
docker-compose build
# avoid docker-compose creating the volume if it doesn't exist
if [ -e "$YOUR_CRATE" ]; then
  docker-compose run -e DOCSRS_DOCKER_IMAGE=build-env \
                     -e RUST_BACKTRACE=1 \
                     -v "$YOUR_CRATE":/opt/rustwide/workdir \
    web build crate --local /opt/rustwide/workdir
else
  echo "$YOUR_CRATE does not exist";
fi

Making multiple changes

If your build fails even after your changes, it will be annoying to rebuild the image from scratch just to add a single package. Instead, you can make changes directly to the Dockerfile so that the existing packages are cached. Be sure to move these new packages from the Dockerfile to packages.txt once you are sure they work.

On line 7 of the Dockerfile, add this line: RUN apt-get install -y your_second_package. Rerun the build and start the container; it should take much less time now:

cd ../crates-build-env/linux
docker build --tag build-env .
cd ../../docs.rs
docker-compose run -e DOCSRS_DOCKER_IMAGE=build-env \
                     -e RUST_BACKTRACE=1 \
                     -v "$YOUR_CRATE":/opt/rustwide/workdir \
    web build crate --local /opt/rustwide/workdir

Run the lint script

Before you make a PR, run the shell script lint.sh and make sure it passes. It ensures packages.txt is in order and will tell you exactly what changes you need to make if not.

cd ../crates-build-env
./lint.sh

Make a pull request

Once you are sure your package builds, you can make a pull request to get it adopted upstream for docs.rs and crater. Go to https://github.com/rust-lang/crates-build-env and click 'Fork' in the top right. Locally, add your fork as a remote in git and push your changes:

git remote add personal https://github.com/<your_username_here>/crates-build-env
git add -u
git commit -m 'add packages necessary for <your_package_here> to compile'
git push personal

Back on github, make a pull request:

1. Go to https://github.com/rust-lang/crates-build-env/compare
2. Click 'compare across forks'
3. Click 'head repository' -> <your_username>/crates-build-env
4. Click 'Create pull request'
5. Add a description of what packages you added and what crate they fixed
6. Click 'Create pull request' again in the bottom right.

Hopefully your changes will be merged quickly! After that you can either publish a point release (rebuilds your docs immediately) or request for a member of the docs.rs team to schedule a new build (may take a while depending on their schedules).

Self hosting a docs.rs instance

These are instructions for deploying the server in a production environment. For instructions on developing locally without docker-compose, see Developing without docker-compose.

Here is a breakdown of what it takes to turn a regular server into its own version of docs.rs.

Beware: This process is rather rough! Attempts at cleaning it up, automating setup components, etc, would be greatly appreciated!

Requirements

The commands and package names on this page will assume an Ubuntu server running systemd, but hopefully the explanatory text should give enough information to adapt to other systems. Note that docs.rs depends on the host being x86_64-unknown-linux-gnu.

Docs.rs has a few basic requirements:

• Rust (preferably via rustup)
• Git
• CMake, GCC, G++, and pkg-config (to build dependencies for crates and docs.rs itself)
• OpenSSL, zlib, curl, and libmagic (to link against)
• PostgreSQL
• LXC tools (doc builds run inside an LXC container)

$ curl https://sh.rustup.rs -sSf | sh -s -- -y --default-toolchain nightly
$ source $HOME/.cargo/env
# apt install build-essential git curl cmake gcc g++ pkg-config libmagic-dev libssl-dev zlib1g-dev postgresql lxc-utils

The cratesfyi user

To help things out later on, we can create a new unprivileged user that will run the server process. This user will own all the files required by the docs.rs process. This user will need to be able to run lxc-attach through sudo to be able to run docs builds, so give it a sudoers file at the same time:

# adduser --disabled-login --disabled-password --gecos "" cratesfyi
# echo 'cratesfyi  ALL=(ALL) NOPASSWD: /usr/bin/lxc-attach' > /etc/sudoers.d/cratesfyi

(The name cratesfyi is a historical one: Before the site was called "docs.rs", it was called "crates.fyi" instead. If you want to update the name of the user, feel free! Just be aware that the name cratesfyi will be used throughout this document.)

The "prefix" directory

In addition to the LXC container, docs.rs also stores several related files in a "prefix" directory. This directory can be stored anywhere, but the cratesfyi user needs to be able to access it:

# mkdir /cratesfyi-prefix
# chown cratesfyi:cratesfyi /cratesfyi-prefix

Now we can set up some required folders. To make sure they all have proper ownership, run them all as cratesfyi:

$ sudo -u cratesfyi mkdir -vp /cratesfyi-prefix/documentations /cratesfyi-prefix/public_html /cratesfyi-prefix/sources
$ sudo -u cratesfyi git clone https://github.com/rust-lang/crates.io-index.git /cratesfyi-prefix/crates.io-index
$ sudo -u cratesfyi git --git-dir=/cratesfyi-prefix/crates.io-index/.git branch crates-index-diff_last-seen

(That last command is used to set up the crates-index-diff crate, so we can start monitoring new crate releases.)

LXC container

To help contain what crates' build scripts can access, documentation builds run inside an LXC container. To create one inside the prefix directory:

# LANG=C lxc-create -n cratesfyi-container -P /cratesfyi-prefix -t download -- --dist ubuntu --release bionic --arch amd64
# ln -s /cratesfyi-prefix/cratesfyi-container /var/lib/lxc
# chmod 755 /cratesfyi-prefix/cratesfyi-container
# chmod 755 /var/lib/lxc

(To make deployment simpler, it's important that the OS the container is using is the same as the host! In this case, the host is assumed to be running 64-bit Ubuntu 18.04. If you make the container use a different release or distribution, you'll need to build docs.rs separately inside the container when deploying.)

You'll also need to configure networking for the container. The following is a sample /etc/default/lxc-net that enables NAT networking for the container:

USE_LXC_BRIDGE="true"
LXC_BRIDGE="lxcbr0"
LXC_ADDR="10.0.3.1"
LXC_NETMASK="255.255.255.0"
LXC_NETWORK="10.0.3.0/24"
LXC_DHCP_RANGE="10.0.3.2,10.0.3.254"
LXC_DHCP_MAX="253"
LXC_DHCP_CONFILE=""
LXC_DOMAIN=""

In addition, you'll need to set the container's configuration to use this. Add the following lines to /cratesfyi-prefix/cratesfyi-container/config:

lxc.net.0.type = veth
lxc.net.0.link = lxcbr0

Now you can reload the LXC network configuration, start up the container, and set it up to auto-start when the host boots:

# systemctl restart lxc-net
# systemctl enable lxc@cratesfyi-container.service
# systemctl start lxc@cratesfyi-container.service

Now we need to do some setup inside this container. You can either copy all these commands so that each one attaches on its own, or you can run lxc-console -n cratesfyi-container to open a root shell inside the container and skip the lxc-attach prefix.

# lxc-attach -n cratesfyi-container -- apt update
# lxc-attach -n cratesfyi-container -- apt upgrade
# lxc-attach -n cratesfyi-container -- apt install curl ca-certificates binutils gcc libc6-dev libmagic1 pkg-config build-essential

Inside the container, we also need to set up a cratesfyi user, and install Rust for it. In addition to the base Rust installation, we also need to install all the default targets so that we can build docs for all the Tier 1 platforms. The Rust compiler installed inside the container is the one that builds all the docs, so if you want to use a new Rustdoc feature, this is the compiler to update.

lxc-attach -n cratesfyi-container -- adduser --disabled-login --disabled-password --gecos "" cratesfyi
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'curl https://sh.rustup.rs -sSf | sh -s -- -y --default-toolchain nightly'
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup target add i686-apple-darwin'
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup target add i686-pc-windows-msvc'
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup target add i686-unknown-linux-gnu'
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup target add x86_64-apple-darwin'
lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup target add x86_64-pc-windows-msvc'

Now that we have Rust installed inside the container, we can use a trick to give the cratesfyi user on the host the same Rust compiler as the container. By symlinking the following directories into its user directory, we don't need to track a third toolchain.

for directory in .cargo .rustup .multirust; do  [[ -h /home/cratesfyi/$directory ]] || sudo -u cratesfyi ln -vs /var/lib/lxc/cratesfyi-container/rootfs/home/cratesfyi/$directory /home/cratesfyi/; done

Environment for the cratesfyi user

To ensure that the docs.rs server is configured properly, we need to set a few environment variables. The primary ones are going into a separate environment file, so we can load them into the systemd service that will manage the server.

Write the following into /home/cratesfyi/.cratesfyi.env. If you have a GitHub access token that the site can use to collect repository information, add it here, but otherwise leave it blank. The variables need to exist, but they can be blank to skip that collection.

CRATESFYI_PREFIX=/cratesfyi-prefix
CRATESFYI_DATABASE_URL=postgresql://cratesfyi:password@localhost
CRATESFYI_CONTAINER_NAME=cratesfyi-container
CRATESFYI_GITHUB_USERNAME=
CRATESFYI_GITHUB_ACCESSTOKEN=
RUST_LOG=cratesfyi

Now add the following to /home/cratesfyi/.profile:

export $(cat $HOME/.cratesfyi.env | xargs -d '\n')
export PATH="$HOME/.cargo/bin:$PATH"
export PATH="$PATH:$HOME/docs.rs/target/release"

Docs.rs build

Now we can actually clone and build the docs.rs source! The location of it doesn't matter much, but again, we want it to be owned by cratesfyi so it can build and run the final executable. In addition, we copy the built cratesfyi binary into the container so that it can be used to arrange builds on the inside.

sudo -u cratesfyi git clone https://github.com/rust-lang-nursery/docs.rs.git ~cratesfyi/docs.rs
sudo su - cratesfyi -c 'cd ~/docs.rs && cargo build --release'
cp -v /home/cratesfyi/docs.rs/target/release/cratesfyi /var/lib/lxc/cratesfyi-container/rootfs/usr/local/bin

PostgreSQL

Now that we have the repository built, we can use it to set up the database. Docs.rs uses a Postgres database to store information about crates and their documentation. To set one up, we first need to ask Postgres to create the database, and then run the docs.rs command to create the initial tables and content:

sudo -u postgres sh -c "psql -c \"CREATE USER cratesfyi WITH PASSWORD 'password';\""
sudo -u postgres sh -c "psql -c \"CREATE DATABASE cratesfyi OWNER cratesfyi;\""
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- database init"
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build add-essential-files"
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build crate rand 0.5.5"
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- database update-search-index"
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- database update-release-activity"

Server configuration

We're almost there! At this point, we've got all the pieces in place to run the site. Now we can set up a systemd service that will run the daemon that will collect crate information, orchestrate builds, and serve the website. The following systemd service file can be placed in /etc/systemd/system/cratesfyi.service:

[Unit]
Description=Cratesfyi daemon
After=network.target postgresql.service

[Service]
User=cratesfyi
Group=cratesfyi
Type=forking
PIDFile=/cratesfyi-prefix/cratesfyi.pid
EnvironmentFile=/home/cratesfyi/.cratesfyi.env
ExecStart=/home/cratesfyi/docs.rs/target/release/cratesfyi daemon
WorkingDirectory=/home/cratesfyi/docs.rs

[Install]
WantedBy=multi-user.target

Enabling and running that will serve the website on http://localhost:3000, so if you want to route public traffic to it, you'll need to set up something like nginx to proxy the connections to it.

Updating Rust

If you want to update the Rust compiler used to build crates (and the Rustdoc that comes with it), you need to make sure you don't interrupt any existing crate builds. The daemon waits for 60 seconds between checking for new crates, so you need to make sure you catch it during that window. Since we hooked the daemon into systemd, the logs will be available in its journal. Running journalctl -efu cratesfyi (it may need to be run as root if nothing appears) will show the latest log output and show new entries as they appear. You're looking for a message like "Finished building new crates, going back to sleep" or "Queue is empty, going back to sleep", which indicates that the crate-building thread is waiting.

To prevent the queue from building more crates, run the following:

sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build lock"

This will create a lock file in the prefix directory that will prevent more crates from being built. At this point, you can update the rustc inside the container and add the rustdoc static files to the database:

lxc-attach -n cratesfyi-container -- su - cratesfyi -c 'rustup update'
sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build add-essential-files"

Once this is done, you can unlock the queue to allow crates to build again:

sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build unlock"

And we're done! New crates will start being built with the new rustc. If you want to rebuild any existing docs with the new rustdoc, you need to manually build them - there's no automated way to rebuild failed docs or docs from a certain rust version yet.

Updating docs.rs

To update the code for docs.rs itself, you can follow a similar approach. First, watch the logs so you can stop the daemon from building more crates. (You can replace the lock command with a systemctl stop cratesfyi if you don't mind the web server being down while you build.)

# journalctl -efu cratesfyi
(wait for build daemon to sleep)
$ sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build lock"

Once the daemon has stopped, you can start updating the code and rebuilding:

$ sudo su - cratesfyi -c "cd ~/docs.rs && git pull"
$ sudo su - cratesfyi -c "cd ~/docs.rs && cargo build --release"

Now that we have a shiny new build, we need to make sure the service is using it:

# cp -v /home/cratesfyi/docs.rs/target/release/cratesfyi /var/lib/lxc/cratesfyi-container/rootfs/usr/local/bin
# systemctl restart cratesfyi

Next, we can unlock the builder so it can start checking new crates:

$ sudo su - cratesfyi -c "cd ~/docs.rs && cargo run --release -- build unlock"

And we're done! Changes to the site or the build behavior should be visible now.

Common maintenance procedures

Temporarily remove a crate from the queue

It might happen that a crate fails to build repeatedly due to a docs.rs bug, clogging up the queue and preventing other crates to build. In this case it's possible to temporarily remove the crate from the queue until the docs.rs's bug is fixed. To do that, log into the machine and open a PostgreSQL shell with:

$ psql

Then you can run this SQL query to remove the crate:

UPDATE queue SET attempt = 100 WHERE name = '<CRATE_NAME>';

To add the crate back in the queue you can run in the PostgreSQL shell this query:

UPDATE queue SET attempt = 0 WHERE name = '<CRATE_NAME>';

Pinning a version of nightly

Sometimes the latest nightly might be broken, causing doc builds to fail. In those cases it's possible to tell docs.rs to stop updating to the latest nightly and instead pin a specific release. To do that you need to edit the /home/cratesfyi/.docs-rs-env file, adding or changing this environment variable:

CRATESFYI_TOOLCHAIN=nightly-YYYY-MM-DD

Once the file changed docs.rs needs to be restarted:

systemctl restart docs.rs

To return to the latest nightly simply remove the environment variable and restart docs.rs again.

Rebuild a specific crate

If a bug was recently fixed, you may want to rebuild a crate so that it builds with the latest version. From the docs.rs machine:

cratesfyi queue add <crate> <version>

This will add the crate with a lower priority than new crates by default, you can change the priority with the -p option.

Raise the limits for a specific crate

Occasionally crates will ask for their build limits to be raised. You can raise them from the docs.rs machine with psql.

Raising a memory limit to 8 GB:

# memory is measured in bytes
cratesfyi=> INSERT INTO sandbox_overrides (crate_name, max_memory_bytes)
  VALUES ('crate name', 8589934592);

Raising a timeout to 15 minutes:

cratesfyi=> INSERT INTO sandbox_overrides (crate_name, timeout_seconds)
  VALUES ('crate name', 900);

Raising limits for multiple crates at once:

cratesfyi=> INSERT INTO sandbox_overrides (crate_name, max_memory_bytes)
  VALUES ('stm32f4', 8589934592), ('stm32h7', 8589934592), ('stm32g4', 8589934592);

Set a group of crates to be automatically de-prioritized

When many crates from the same project are published at once, they take up a lot of space in the queue. You can de-prioritize groups of crates at once like this:

cratesfyi=> INSERT INTO crate_priorities (pattern, priority)
  VALUES ('group-%', 1);

The pattern should be a LIKE pattern as documented on https://www.postgresql.org/docs/current/functions-matching.html.

Note that this only sets the default priority for crates with that name. If there are crates already in the queue, you'll have to update those manually:

cratesfyi=> UPDATE queue SET priority = 1 WHERE name LIKE 'group-%';

Adding all the crates failed after a date back in the queue

After an outage you might want to add all the failed builds back to the queue. To do that, log into the machine and open a PostgreSQL shell with:

psql

Then you can run this SQL query to add all the crates failed after YYYY-MM-DD HH:MM:SS back in the queue:

UPDATE queue SET attempt = 0 WHERE attempt >= 5 AND build_time > 'YYYY-MM-DD HH:MM:SS';

Removing a crate from the website

Sometimes it might be needed to remove all the content related to a crate from docs.rs (for example after receiving a DMCA). To do that, log into the server and run:

cratesfyi database delete-crate CRATE_NAME

The command will remove all the data from the database, and then remove the files from S3.

Blacklisting crates

Occasionally it might be needed to prevent a crate from being built on docs.rs, for example if we can't legally host the content of those crates. To add a crate to the blacklist, preventing new builds for it, you can run:

cratesfyi database blacklist add <CRATE_NAME>

Other operations (such as list and remove) are also supported.

Warning: blacklisting a crate doesn't remove existing content from the website, it just prevents new versions from being built!

Governance

IMPORTANT This document is adapted from RFC 1068 and is currently being actively worked on, however there may be large parts of Rust's governance that are missing, incomplete, or out of date.

Core team

The core team serves as leadership for the Rust project as a whole. In particular, it:

• Sets the overall direction and vision for the project. That means setting the core values that are used when making decisions about technical tradeoffs. It means steering the project toward specific use cases where Rust can have a major impact. It means leading the discussion, and writing RFCs for, major initiatives in the project.

• Sets the priorities and release schedule. Design bandwidth is limited, and it's dangerous to try to grow the language too quickly; the core team makes some difficult decisions about which areas to prioritize for new design, based on the core values and target use cases.

• Focuses on broad, cross-cutting concerns. The core team is specifically designed to take a global view of the project, to make sure the pieces are fitting together in a coherent way.

• Spins up or shuts down subteams. Over time, we may want to expand the set of subteams, and it may make sense to have temporary "strike teams" that focus on a particular, limited task.

• Decides whether/when to ungate a feature. While the subteams make decisions on RFCs, the core team is responsible for pulling the trigger that moves a feature from nightly to stable. This provides an extra check that features have adequately addressed cross-cutting concerns, that the implementation quality is high enough, and that language/library commitments are reasonable.

The core team should include both the subteam leaders, and, over time, a diverse set of other stakeholders that are both actively involved in the Rust community, and can speak to the needs of major Rust constituencies, to ensure that the project is addressing real-world needs.

Subteams

The primary roles of each subteam are:

• Shepherding RFCs for the subteam area. As always, that means (1) ensuring that stakeholders are aware of the RFC, (2) working to tease out various design tradeoffs and alternatives, and (3) helping build consensus.

• Accepting or rejecting RFCs in the subteam area.

• Setting policy on what changes in the subteam area require RFCs, and reviewing direct PRs for changes that do not require an RFC.

• Delegating reviewer rights for the subteam area. The ability to r+ is not limited to team members, and in fact earning r+ rights is a good stepping stone toward team membership. Each team should set reviewing policy, manage reviewing rights, and ensure that reviews take place in a timely manner. (Thanks to Nick Cameron for this suggestion.)

Subteams make it possible to involve a larger, more diverse group in the decision-making process. In particular, they should involve a mix of:

• Rust project leadership, in the form of at least one core team member (the leader of the subteam).

• Area experts: people who have a lot of interest and expertise in the subteam area, but who may be far less engaged with other areas of the project.

• Stakeholders: people who are strongly affected by decisions in the subteam area, but who may not be experts in the design or implementation of that area. It is crucial that some people heavily using Rust for applications/libraries have a seat at the table, to make sure we are actually addressing real-world needs.

Members should have demonstrated a good sense for design and dealing with tradeoffs, an ability to work within a framework of consensus, and of course sufficient knowledge about or experience with the subteam area. Leaders should in addition have demonstrated exceptional communication, design, and people skills. They must be able to work with a diverse group of people and help lead it toward consensus and execution.

Each subteam is led by a member of the core team. The leader is responsible for:

• Setting up the subteam:

  • Deciding on the initial membership of the subteam (in consultation with the core team). Once the subteam is up and running.

  • Working with subteam members to determine and publish subteam policies and mechanics, including the way that subteam members join or leave the team (which should be based on subteam consensus).

• Communicating core team vision downward to the subteam.

• Alerting the core team to subteam RFCs that need global, cross-cutting attention, and to RFCs that have entered the "final comment period" (see below).

• Ensuring that RFCs and PRs are progressing at a reasonable rate, re-assigning shepherds/reviewers as needed.

• Making final decisions in cases of contentious RFCs that are unable to reach consensus otherwise (should be rare).

The way that subteams communicate internally and externally is left to each subteam to decide, but:

• Technical discussion should take place as much as possible on public forums, ideally on RFC/PR threads and tagged discuss posts.

• Each subteam will have a dedicated internals forum tag.

• Subteams should actively seek out discussion and input from stakeholders who are not members of the team.

• Subteams should have some kind of regular meeting or other way of making decisions. The content of this meeting should be summarized with the rationale for each decision -- and, as explained below, decisions should generally be about weighting a set of already-known tradeoffs, not discussing or discovering new rationale.

• Subteams should regularly publish the status of RFCs, PRs, and other news related to their area. Ideally, this would be done in part via a dashboard like the Homu queue.

Decision-making

Consensus

Rust has long used a form of consensus decision-making. In a nutshell the premise is that a successful outcome is not where one side of a debate has "won", but rather where concerns from all sides have been addressed in some way. This emphatically does not entail design by committee, nor compromised design. Rather, it's a recognition that

... every design or implementation choice carries a trade-off and numerous costs. There is seldom a right answer.

Breakthrough designs sometimes end up changing the playing field by eliminating tradeoffs altogether, but more often difficult decisions have to be made. The key is to have a clear vision and set of values and priorities, which is the core team's responsibility to set and communicate, and the subteam's responsibility to act upon.

Whenever possible, we seek to reach consensus through discussion and design revision. Concretely, the steps are:

• Initial RFC proposed, with initial analysis of tradeoffs.
• Comments reveal additional drawbacks, problems, or tradeoffs.
• RFC revised to address comments, often by improving the design.
• Repeat above until "major objections" are fully addressed, or it's clear that there is a fundamental choice to be made.

Consensus is reached when most people are left with only "minor" objections, i.e., while they might choose the tradeoffs slightly differently they do not feel a strong need to actively block the RFC from progressing.

One important question is: consensus among which people, exactly? Of course, the broader the consensus, the better. But at the very least, consensus within the members of the subteam should be the norm for most decisions. If the core team has done its job of communicating the values and priorities, it should be possible to fit the debate about the RFC into that framework and reach a fairly clear outcome.

Lack of consensus

In some cases, though, consensus cannot be reached. These cases tend to split into two very different camps:

• "Trivial" reasons, e.g., there is not widespread agreement about naming, but there is consensus about the substance.

• "Deep" reasons, e.g., the design fundamentally improves one set of concerns at the expense of another, and people on both sides feel strongly about it.

In either case, an alternative form of decision-making is needed.

• For the "trivial" case, usually either the RFC shepherd or subteam leader will make an executive decision.

• For the "deep" case, the subteam leader is empowered to make a final decision, but should consult with the rest of the core team before doing so.

How and when RFC decisions are made, and the "final comment period"

Each RFC has a shepherd drawn from the relevant subteam. The shepherd is responsible for driving the consensus process -- working with both the RFC author and the broader community to dig out problems, alternatives, and improved design, always working to reach broader consensus.

At some point, the RFC comments will reach a kind of "steady state", where no new tradeoffs are being discovered, and either objections have been addressed, or it's clear that the design has fundamental downsides that need to be weighed.

At that point, the shepherd will announce that the RFC is in a "final comment period" (which lasts for one week). This is a kind of "last call" for strong objections to the RFC. The announcement of the final comment period for an RFC should be very visible; it should be included in the subteam's periodic communications.

Note that the final comment period is in part intended to help keep RFCs moving. Historically, RFCs sometimes stall out at a point where discussion has died down but a decision isn't needed urgently. In this proposed model, the RFC author could ask the shepherd to move to the final comment period (and hence toward a decision).

After the final comment period, the subteam can make a decision on the RFC. The role of the subteam at that point is not to reveal any new technical issues or arguments; if these come up during discussion, they should be added as comments to the RFC, and it should undergo another final comment period.

Instead, the subteam decision is based on weighing the already-revealed tradeoffs against the project's priorities and values (which the core team is responsible for setting, globally). In the end, these decisions are about how to weight tradeoffs. The decision should be communicated in these terms, pointing out the tradeoffs that were raised and explaining how they were weighted, and never introducing new arguments.

Keeping things lightweight

In addition to the "final comment period" proposed above, this RFC proposes some further adjustments to the RFC process to keep it lightweight.

A key observation is that, thanks to the stability system and nightly/stable distinction, it's easy to experiment with features without commitment.

Clarifying what needs an RFC

Over time, we've been drifting toward requiring an RFC for essentially any user-facing change, which sometimes means that very minor changes get stuck awaiting an RFC decision. While subteams + final comment period should help keep the pipeline flowing a bit better, it would also be good to allow "minor" changes to go through without an RFC, provided there is sufficient review in some other way. (And in the end, the core team ungates features, which ensures at least a final review.)

This RFC does not attempt to answer the question "What needs an RFC", because that question will vary for each subteam. However, this RFC stipulates that each subteam should set an explicit policy about:

1. What requires an RFC for the subteam's area, and
2. What the non-RFC review process is.

These guidelines should try to keep the process lightweight for minor changes.

Clarifying the "finality" of RFCs

While RFCs are very important, they do not represent the final state of a design. Often new issues or improvements arise during implementation, or after gaining some experience with a feature. The nightly/stable distinction exists in part to allow for such design iteration.

Thus RFCs do not need to be "perfect" before acceptance. If consensus is reached on major points, the minor details can be left to implementation and revision.

Later, if an implementation differs from the RFC in substantial ways, the subteam should be alerted, and may ask for an explicit amendment RFC. Otherwise, the changes should just be explained in the commit/PR.

The teams

With all of that out of the way, what subteams should we start with? This RFC proposes the following initial set:

• Language design
• Libraries
• Compiler
• Tooling and infrastructure
• Moderation

In the long run, we will likely also want teams for documentation and for community events, but these can be spun up once there is a more clear need (and available resources).

Language design team

Focuses on the design of language-level features; not all team members need to have extensive implementation experience.

Some example RFCs that fall into this area:

• Associated types and multidispatch
• DST coercions
• Trait-based exception handling
• Rebalancing coherence
• Integer overflow (this has high overlap with the library subteam)
• Sound generic drop

Library team

Oversees both std and, ultimately, other crates in the rust-lang github organization. The focus up to this point has been the standard library, but we will want "official" libraries that aren't quite std territory but are still vital for Rust. (The precise plan here, as well as the long-term plan for std, is one of the first important areas of debate for the subteam.) Also includes API conventions.

Some example RFCs that fall into this area:

• Collections reform
• IO reform
• Debug improvements
• Simplifying std::hash
• Conventions for ownership variants

Compiler team

Focuses on compiler internals, including implementation of language features. This broad category includes work in codegen, factoring of compiler data structures, type inference, borrowck, and so on.

There is a more limited set of example RFCs for this subteam, in part because we haven't generally required RFCs for this kind of internals work, but here are two:

• Non-zeroing dynamic drops (this has high overlap with language design)
• Incremental compilation

Tooling and infrastructure team

Even more broad is the "tooling" subteam, which at inception is planned to encompass every "official" (rust-lang managed) non-rustc tool:

• rustdoc
• rustfmt
• Cargo
• crates.io
• CI infrastructure
• Debugging tools
• Profiling tools
• Editor/IDE integration
• Refactoring tools

It's not presently clear exactly what tools will end up under this umbrella, nor which should be prioritized.

Moderation team

Finally, the moderation team is responsible for dealing with CoC violations.

One key difference from the other subteams is that the moderation team does not have a leader. Its members are chosen directly by the core team, and should be community members who have demonstrated the highest standard of discourse and maturity. To limit conflicts of interest, the moderation subteam should not include any core team members. However, the subteam is free to consult with the core team as it deems appropriate.

The moderation team will have a public email address that can be used to raise complaints about CoC violations (forwards to all active moderators).

Initial plan for moderation

What follows is an initial proposal for the mechanics of moderation. The moderation subteam may choose to revise this proposal by drafting an RFC, which will be approved by the core team.

Moderation begins whenever a moderator becomes aware of a CoC problem, either through a complaint or by observing it directly. In general, the enforcement steps are as follows:

These steps are adapted from text written by Manish Goregaokar, who helped articulate them from experience as a Stack Exchange moderator.

• Except for extreme cases (see below), try first to address the problem with a light public comment on thread, aimed to de-escalate the situation. These comments should strive for as much empathy as possible. Moderators should emphasize that dissenting opinions are valued, and strive to ensure that the technical points are heard even as they work to cool things down.

  When a discussion has just gotten a bit heated, the comment can just be a reminder to be respectful and that there is rarely a clear "right" answer. In cases that are more clearly over the line into personal attacks, it can directly call out a problematic comment.

• If the problem persists on thread, or if a particular person repeatedly comes close to or steps over the line of a CoC violation, moderators then email the offender privately. The message should include relevant portions of the CoC together with the offending comments. Again, the goal is to de-escalate, and the email should be written in a dispassionate and empathetic way. However, the message should also make clear that continued violations may result in a ban.

• If problems still persist, the moderators can ban the offender. Banning should occur for progressively longer periods, for example starting at 1 day, then 1 week, then permanent. The moderation subteam will determine the precise guidelines here.

In general, moderators can and should unilaterally take the first step, but steps beyond that (particularly banning) should be done via consensus with the other moderators. Permanent bans require core team approval.

Some situations call for more immediate, drastic measures: deeply inappropriate comments, harassment, or comments that make people feel unsafe. (See the code of conduct for some more details about this kind of comment). In these cases, an individual moderator is free to take immediate, unilateral steps including redacting or removing comments, or instituting a short-term ban until the subteam can convene to deal with the situation.

The moderation team is responsible for interpreting the CoC. Drastic measures like bans should only be used in cases of clear, repeated violations.

Moderators themselves are held to a very high standard of behavior, and should strive for professional and impersonal interactions when dealing with a CoC violation. They should always push to de-escalate. And they should recuse themselves from moderation in threads where they are actively participating in the technical debate or otherwise have a conflict of interest. Moderators who fail to keep up this standard, or who abuse the moderation process, may be removed by the core team.

Subteam, and especially core team members are also held to a high standard of behavior. Part of the reason to separate the moderation subteam is to ensure that CoC violations by Rust's leadership be addressed through the same independent body of moderators.

Moderation covers all rust-lang venues, which currently include github repos, IRC channels (#rust, #rust-internals, #rustc, #rust-libs), and the two discourse forums. (The subreddit already has its own moderation structure, and isn't directly associated with the rust-lang organization.)

Infrastructure

This section documents Rust's infrastructure, and how it is maintained.

External Links

• rust-toolstate records build and test status of external tools bundled with the Rust repository.

Other Rust Installation Methods

• Which installer should you use?
• Other ways to install rustup
• Standalone installers
• Source code

Which installer should you use?

Rust runs on many platforms, and there are many ways to install Rust. If you want to install Rust in the most straightforward, recommended way, then follow the instructions on the main installation page.

That page describes installation via rustup, a tool that manages multiple Rust toolchains in a consistent way across all platforms Rust supports. Why might one not want to install using those instructions?

• Offline installation. rustup downloads components from the internet on demand. If you need to install Rust without access to the internet, rustup is not suitable.
• Preference for the system package manager. On Linux in particular, but also on macOS with Homebrew, MacPorts or pkgsrc, and Windows with Chocolatey or Scoop, developers sometimes prefer to install Rust with their platform's package manager.
• Preference against curl | sh. On Unix, we usually install rustup by running a shell script via curl. Some have concerns about the security of this arrangement and would prefer to download and run the installer themselves.
• Validating signatures. Although rustup performs its downloads over HTTPS, the only way to verify the signatures of Rust installers today is to do so manually with the standalone installers.
• GUI installation and integration with "Add/Remove Programs" on Windows. rustup runs in the console and does not register its installation like typical Windows programs. If you prefer a more typical GUI installation on Windows there are standalone .msi installers. In the future rustup will also have a GUI installer on Windows.

Rust's platform support is defined in three tiers, which correspond closely with the installation methods available: in general, the Rust project provides binary builds for all tier 1 and tier 2 platforms, and they are all installable via rustup. Some tier 2 platforms though have only the standard library available, not the compiler itself; that is, they are cross-compilation targets only; Rust code can run on those platforms, but they do not run the compiler itself. Such targets can be installed with the rustup target add command.

Other ways to install rustup

The way to install rustup differs by platform:

• On Unix, run curl https://sh.rustup.rs -sSf | sh in your shell. This downloads and runs rustup-init.sh, which in turn downloads and runs the correct version of the rustup-init executable for your platform.
• On Windows, download and run rustup-init.exe.

rustup-init can be configured interactively, and all options can additionally be controlled by command-line arguments, which can be passed through the shell script. Pass --help to rustup-init as follows to display the arguments rustup-init accepts:

curl https://sh.rustup.rs -sSf | sh -s -- --help

If you prefer not to use the shell script, you may directly download rustup-init for the platform of your choice:

• aarch64-linux-android
• aarch64-unknown-linux-gnu
• aarch64-unknown-linux-musl
• arm-linux-androideabi
• arm-unknown-linux-gnueabi
• arm-unknown-linux-gnueabihf
• armv7-linux-androideabi
• armv7-unknown-linux-gnueabihf
• i686-apple-darwin
• i686-linux-android
• i686-pc-windows-gnu
• i686-pc-windows-msvc
• i686-unknown-linux-gnu
• mips-unknown-linux-gnu
• mips64-unknown-linux-gnuabi64
• mips64el-unknown-linux-gnuabi64
• mipsel-unknown-linux-gnu
• powerpc-unknown-linux-gnu
• powerpc64-unknown-linux-gnu
• powerpc64le-unknown-linux-gnu
• s390x-unknown-linux-gnu
• x86_64-apple-darwin
• x86_64-linux-android
• x86_64-pc-windows-gnu
• x86_64-pc-windows-msvc
• x86_64-unknown-freebsd
• x86_64-unknown-illumos
• x86_64-unknown-linux-gnu
• x86_64-unknown-linux-musl
• x86_64-unknown-netbsd

Standalone installers

The official Rust standalone installers contain a single release of Rust, and are suitable for offline installation. They come in three forms: tarballs (extension .tar.gz), that work in any Unix-like environment, Windows installers (.msi), and Mac installers (.pkg). These installers come with rustc, cargo, rustdoc, the standard library, and the standard documentation, but do not provide access to additional cross-targets like rustup does.

The most common reasons to use these are:

• Offline installation
• Preferring a more platform-integrated, graphical installer on Windows

Each of these binaries is signed with the Rust signing key, which is available on keybase.io, by the Rust build infrastructure, with GPG. In the tables below, the .asc files are the signatures.

Source code

The Rust Release Channel Layout

NOTE This document should be considered incomplete and descriptive rather than normative. Do not rely on anything described herein to be fully correct or a definition of how things should be done.

A lot of the content herein is derived from a posting made to the Rust internals forum by Brian Anderson back in 2016.

Rust releases are deployed onto static.rust-lang.org where they are served via https. There are several parts to a release channel (stable, beta, nightly) but they all key off a manifest file and then go from there.

Channel manifests

There is a top level directory /dist/ which contains the channel manifests. The manifests are named channel-rust-[channelname].toml. Each channel manifest is accompanied by a .sha256 file which is a checksum of the manifest file and can be used to check integrity of the downloaded data. In addition each channel's manifest is also accompanied by a .asc file which is a detached GPG signature which can be used to check not only the integrity but also the authenticity of the channel manifest.

In addition to the stable, beta, and nightly channels, there is also a manifest for each release which will be called channel-rust-x.yy.z.toml with its associated .sha256 and .asc files.

To support date-based channels, there is an archive folder for each day (labelled YYYY-MM-DD) which contains copies of the requisite channel files on that day. So, for example, if you installed nightly-2019-02-16 then the channel file would be https://static.rust-lang.org/dist/2019-02-16/channel-rust-nightly.toml.

Content of channel manifests

Channel manifests are toml files. These are known as v2 manifests. The v1 manifests are simply lists of the files associated with a release and are not generated for every channel all of the time. Currently it is recommended to work only with the v2 manifests and these are the topic of this section.

The top level of the .toml file consists of two important key/value pairs. Firstly the manifest-version which is, at this time, "2", and secondly the date of the manifest (date) whose value is of the form "YYYY-MM-DD".

There are then a number of top level sections (tables) which are:

• pkg - This contains the bulk of the manifest and lists the packages which are part of the release. Typically this will be things like rust, rustc, cargo etc. The rust package is semi-special and currently is used to specify the subset of other packages which will be installed by default.

  Within packages are components and extensions. Currently components are installed by default by rustup, extensions are optional components and are available via rustup component add and friends.

• renames - This contains a set of package renames which can be used to determine the correct package to fetch when the user enters an alias for it.

  Typically renames are used when a package leaves its preview state and is considered to be release quality. For example, the actual package for rustfmt is called rustfmt-preview but since its release there has been a renames.rustfmt table whose to field is rustfmt-preview. When the user runs rustup component add rustfmt the name is automatically translated to rustfmt-preview and when the user runs rustup component list then rustfmt-preview is automatically renamed back to rustfmt for display to the user.

• profiles - This is part of the future setup for deciding the default component set to install. Instead of choosing the components of pkg.rust instead rustup will honor one of the entries in the profiles table. Usually this will be the default entry which essentially (though not exactly) boils down to ["rustc", "cargo", "rust-std", "rust-docs", "rustfmt", "clippy"].

  Other profiles include minimal (["rustc", "cargo", "rust-std"]) and complete which adds in additional things such as a copy of the standard library source (rust-src), miri, lldb, llvm-tools, and rust-analysis.

Package entries in the channel manifest

As stated above, packages list their components and extensions (mostly just the rust package) and they can provide per-target tarball and sha256 data.

For example, a package might be:

[pkg.cargo.target.powerpc64-unknown-linux-gnu]
available = true
url = "https://static.rust-lang.org/dist/2019-05-23/cargo-0.36.0-powerpc64-unknown-linux-gnu.tar.gz"
hash = "279f3a84f40e3547a8532c64643f38068accb91c21f04cd16e46579c893f5a06"
xz_url = "https://static.rust-lang.org/dist/2019-05-23/cargo-0.36.0-powerpc64-unknown-linux-gnu.tar.xz"
xz_hash = "cf93b387508f4aea4e64f8b4887d70cc07a00906b981dc0c143e92e918682e4a"

Here you can see that this is for the cargo package, and for the powerpc64-unknown-linux-gnu target. The url/hash combo is for a .tar.gz and the xz_url/xz_hash pair for the same tarball compressed with xz. Either pair of url and hash could be present, both may be present, but it is not useful for neither to be present unless available is set to false to indicate that that particular combination of package and target is unavailable in this channel at this time.

In addition, there will be a single entry providing the version for a package in the form:

[pkg.cargo]
version = "0.36.0 (6f3e9c367 2019-04-04)"

Here version will be effectively the $tool --version output, minus the tool's name.

Targets

Targets are the same triples you might use when building something with cargo build --target=$target and you can add them to your installation using rustup target add $target. When you do that, what rustup actually does is to find the rust-std package for the target in question and installs that. Essentially like an imaginary rustup component add rust-std.$target.

If a rust-std package for a target is not available = true then that target cannot be installed via rustup. This can happen for lower tier targets from time to time.

Since components and extensions are target-specific in the pkg tables, you will be able to see that rust-std for every target is specified in every rust target's extensions. This allows for cross-compilation by installation of any rust-std on any build system.

Service Infrastructure

Most services in the Rust Infrastructure are deployed via rust-central-station. Questions about infrastructure, including current status, should go to the #t-infra Zulip stream.

Our stability guarantees: many of our services rely on publicly-accessible storage and APIs, but not all of these are intended for public consumption. At the moment, only the resources behind static.rust-lang.org are considered stable, meaning that those resources will not change without (at least) prior notice. If you are relying on other parts of the Rust project infrastructure for your own work, please let the infrastructure team know.

Highfive

Highfive is a bot (bot user account) which welcomes newcomers and assigns reviewers.

Note: Highfive is currently being replaced by rustbot. This service will be shut down in the future once the migration is complete.

Rust Log Analyzer

The Rust Log Analyzer analyzes CI build logs to extract error messages and posts them to the pull request. It is run by TimNN.

Homu / bors

Homu is a bot which manages pull requests. It is often referred to as "bors" due to the name of its bot user account. Approved pull requests are placed in a queue from which tests are run.

Documentation on homu commands can be found here.

Please contact Alex Crichton if something goes wrong with the bot.

rfcbot

rfcbot is a bot (bot user account) which helps manage async decision making on issues and PRs (typically RFCs). Team members can view any pending requests for review on the FCP dashboard.

Documentation on rfcbot commands can be found in the rfcbot repository.

rustbot

rustbot is a bot (bot user account) to assist with managing issues and PRs to allow users to label and assign without GitHub permissions. See the wiki for more information.

DXR

DXR is a cross-referenced source index for Rust, allowing the Rust source tree to be navigated and searched with ease. It is generated by rust-dxr

perf / rust-timer

perf offers information about the performance of rustc over time, and a bot for on-demand benchmarking.

It is split into a data collector and a web frontend + bot. The raw performance data is available here and can be browsed on the perf website.

One-off performance runs can done by addressing the rust-timer bot (bot user account). You can trigger the necessary try-build and queue a perf run by saying

@bors try @rust-timer queue

(Technically, the requirement is that the queue command finishes executing prior to the try build completing successfully.)

See the documentation for further bot commands.

Rust Playground

Rust Playground allows you to experiment with Rust before you install it locally, or in any other case where you might not have the compiler available. The Rust playground can be accessed here.

Crater

Crater is a tool to run experiments across the whole Rust ecosystem. Its primary purpose is to detect regressions in the Rust compiler, and it does this by building large number of crates, running their test suites and comparing the results between two versions of the Rust compiler.

Crates comes with a bot to trigger experiments.

docs.rs

docs.rs builds and serves the rustdoc documentation for all crates on crates.io. Issues may be filed on the docs.rs repository. See the #docs-rs channel on Discord for discussion or urgent issues.

Toolstate

The state of tools included with Rust are tracked on the toolstate page. When each PR is merged via CI, the status of each tool is recorded in a JSON file and stored in the toolstate repo. For further information, see the toolstate system documentation.

Rustup components history

The rustup components history tracks the status of every rustup component for every platform over time. See the repository for more information.

CI Timing Tracker

The CI Timing Tracker tracks and compares how long CI jobs take over time. It is run by Alex Crichton.

Team Maintenance

The roster of the Rust teams is always in flux. From time to time, new people are added, but also people sometimes opt to into "alumni status", meaning that they are not currently an active part of the decision-making process. Unfortunately, whenever a new person is added or someone goes into alumni status, there are a number of disparate places that need to be updated. This page aims to document that list. If you have any questions, or need someone with more privileges to make a change for you, a good place to ask is #infra on Discord.

Team repo

Membership of teams is primarily driven by the config files in the rust-lang/team repo. Several systems use the team repo data to control access:

• the team website
• bors r+ rights
• rfcbot interaction
• Mailgun email lists

Team membership is duplicated in a few other places listed below, but the long-term goal is to centralize on the team repo.

Full team membership

To make a full team member, the following places need to be modified:

• the team repo
• the rust-lang/TEAM and (in some cases) rust-lang-nursery/TEAM teams on github must be updated
• the internals discussion board has per-team groups
• if the member is going to join the review rotation, they will need to be added to the [assign.owners] section of triagebot.toml of the repos where they will be reviewing

Team member departure

Remove the team member from any and all places:

• highfive
• 1password
• The GitHub team, GitHub nursery team
• team repo
• toolstate notifications
• triagebot.toml files of all repos they were involved in

Handling of tools embedded in the rustc repo ("toolstate")

The Rust repository contains several external git submodules (e.g. the Book, the Reference). The toolstate system is used to allow these submodules to be in a broken state, except for beta releases.

This is necessary because the documentation is tested both on the rust-lang/rust CI, and on the CI of the documentation repo. If there is a change to rustc that breaks the documentation, it would not be possible to update the documentation since the not-yet-merged version of rustc that breaks it doesn't exist, yet. We usually require CI to be in a passing state in both repos.

The toolstate system solves this problem by temporarily allowing the documentation to be in a "failing" state on rust-lang/rust. When the tests start failing, the maintainers of the submodule will be notified. They will then be responsible for getting it fixed.

The three possible states of a "tool" are: test-pass, test-fail, build-fail.

This page gives a rough overview how the toolstate system works, and what the rules are for when which tools are (not) allowed to break.

Note: Historically, the toolstate system was used for managing tools that were closely coupled with the compiler (like rustfmt or miri). However, those have since been transitioned to use git subtrees instead, so that those tools must always pass their tests, and any failures must be resolved within the PR that breaks them.

This document uses the term "tool", but as of this writing, the only thing tracked is external documentation.

Toolstate Rules

• For all tools, if a PR changes that tool (if it changes the commit used by the submodule), the tool has to be in test-pass after this PR or else CI will fail.

• For all tools except for "nightly only" tools, the following extra rules are applied:

  • If a PR lands on the beta or stable branch, the tool has to be test-pass.
  • If a PR lands on master in the week before the beta is cut, and that PR regresses the tool (if it makes the state "worse"), CI fails. This is to help make sure all these tools become test-pass so that a beta can be cut. (See the Forge index for when the next beta cutoff is happening.)

  At the time of writing, the following tools are "nightly only": embedded-book.

Updating the toolstate repository

Updating the toolstate repository happens in two steps: when CI runs on the auto branch (where bors moves a PR to test if it is good for integration), the "tool" runners for the individual platforms (at the time of writing, Linux and Windows) each submit a JSON file to the repository recording the state of each tool for the commit they are testing. Later, if that commit actually entirely passed CI and bors moves it to the master branch, the "current tool status" in the toolstate repository is updated appropriately.

These scripts also automatically ping some people and create issues when tools break.

For further details, see the comments in the involved files: checktools.sh, publish_toolstate.py as well as the other files mentioned there.

Updating tools

Tools can be updated by updating the submodule to the proper commit.

Run git submodule update --remote path/to/submodule, add the updates, make sure the tests pass, commit, and send a pull request. The path is from the root of the rust repository, so for example, the reference is src/doc/reference.

While not required, subup may assist you with this.

Adding a tool

NOTE: We are trying to switch away from submodules and toolstate over time. Consider adding a subtree instead of a submodule: #70651

To add a new tool to be tracked, the following steps must be taken:

1. Create a PR to rust-lang/rust that adds the submodule along with any necessary build system / bootstrap updates. Be careful that the tests properly support ./x.py --no-fail-fast to avoid issues like this.
2. Include changes to checktools.sh:
   • Build the tool at the top. This is the step that actually generates the JSON status for the tool. When save-toolstates is set in config.toml, the rust build system will write a JSON file with the status of each test.
   • Add the tool to status_check with whether it should be a beta blocker or not.
3. Update publish_toolstate.py to add the tool. This includes a list of people to ping if the tool is broken, and its source repo. (Note: At the time of this writing, these users must have permissions to be assignable on rust-lang/rust GitHub.)
4. Submit a PR to the toolstate repository to manually add the tool to the latest.json file.

Policies of the infrastructure team

This section documents the policies created by the infrastructure team.

Policy on broken nightlies

Sometimes the nightlies released automatically by our CI ends up being broken for some people or even everyone. This policy defines what the infra team response will be in those cases.

Which nightly will be rolled back

A nightly can only be rolled back in the following cases:

• If it contains destructive code, for example if the included compiler deletes all the users files.
• If an infra problem caused it to be broken for a big percentage of users on any Tier 1 platform. Issues affecting only lower tier platforms are not worthy of a roll back, since we don't guarantee working builds for those platforms anyway.

A nightly will not be rolled back if it's broken by a critical compiler bug: those bugs are supposed to be caught by CI, and nightly can have compiler regressions anyway. There are no exceptions, even if big projects are broken because of this.

What are we going to fix

Once any member of the infra team decides to roll back a nightly under this policy we will roll back to the most recent working nightly. The roll back has to fix installing the nightly with rustup:

$ rustup toolchain install nightly

It's not required to roll back other things like the documentation or the manually downloadable artifacts. After the nightly is rolled back we have to announce the roll back on the @rustlang twitter account and on the status page.

Infrastructure guidelines

This section contains the guidelines written by the infrastructure team for other teams who want to use the project's infrastructure.

Rust Infrastructure hosting for static websites

The Rust Infrastructure team provides hosting for static websites available for all Rust teams. This document explains the requirements a website needs to meet and how to setup one.

Requirements for hosting websites

• The website must be managed by a Rust team, or be officially affiliated with the project.
  The infrastructure team has finite resources and we can't offer hosting for community projects.
• The website’s content and build tooling must be hosted on a GitHub repository in either the rust-lang or rust-lang-nursery organizations.
  The infrastructure team must be able to rebuild the website content at any time (for example if we need to switch hosting), and having it hosted on a GitHub repository inside infra-managed organizations is the best way for us to ensure that. Even though we'd prefer for all the repositories to be public it's not a requirement.
• The website must be built and deployed with a CI service.
  We have custom tooling built around hosting static websites on our infra, and at the moment they work with Travis CI and Azure Pipelines. If you need different CI services ask us in advance and we'll adapt the tooling to your provider of choice.
• The website must reach an A+ grade on the Mozilla Observatory.
  Browsers have multiple security features toggleable only through HTTP response headers, and those features enhance users' privacy and prevent exploits from working. An A+ grade on the Observatory indicates all the important headers are correctly set.
• The website must be hosted on platforms vetted by the infra team.
  We recommend either GitHub Pages or Amazon S3 (in the rust-lang AWS account) as the hosting and CloudFront as the CDN, but if you need other platforms that's good as long as we consider them secure and reliable.

Static websites configuration

To avoid limitations of some hosting providers we have setup CloudFront to enable additional, custom behaviors. These behaviors are configured through a file named website_config.json at the root of the generated website content.

Adding custom headers

One of the requirements for having a static website hosted by the infrastructure team is to reach an A+ grade on the Mozilla Observatory, and that requires custom headers to be set. To setup custom headers you need to add an headers section to website_config.json. This example content includes all the headers needed to reach grade B on the Observatory (to reach grade A+ a Content Security Policy is required):

{
    "headers": {
        "Strict-Transport-Security": "max-age=63072000",
        "X-Content-Type-Options": "nosniff",
        "X-Frame-Options": "DENY",
        "X-XSS-Protection": "1; mode=block",
        "Referrer-Policy": "no-referrer, strict-origin-when-cross-origin"
    }
}

Fixing GitHub Pages redirects

GitHub Pages behaves weirdly when it sits behind CloudFront and it needs to issue redirects: since it doesn't know the real domain name it will use http://org-name.github.io/repo-name as the base of the redirect instead of the correct protocol and domain. To prevent this behavior the github_pages_origin key needs to be added to website_config.json with the origin base url as the value (excluding the protocol):

{
    "github_pages_origin": "org-name.github.io/repo-name"
}

Deployment guide

These deployments steps are meant to be executed by a member of the infrastructure team since they require access to our AWS account.

Configuring AWS

Create a CloudFront web distribution and set the following properties:

• Origin Domain Name: rust-lang.github.io/repo-name
• Origin Protocol Policy: HTTPS Only
• Viewer Protocol Policy: Redirect HTTP to HTTPS
• Lambda Function Association:
  • Viewer Response: arn:aws:lambda:us-east-1:890664054962:function:static-websites:4
• Alternate Domain Names: your-subdomain-name.rust-lang.org
• SSL Certificate: Custom SSL Certificate
  • You will need to request the certificate for that subdomain name through ACM (please use the DNS challenge to validate the certificate)
• Comment: your-subdomain-name.rust-lang.org

Wait until the distribution is propagated and take note of its .cloudfront.net domain name.

Head over to the domain’s Route 53 hosted zone and create a new record set:

• Name: your-subdomain-name
• Type: CNAME
• Value: the .cloudfront.net domain name you saw earlier

Create an AWS IAM user to allow the CI provider used to deploy website changes to perform whitelisted automatic actions. Use ci--ORG-NAME--REPO-NAME (for example ci--rust-lang--rust) as the user name, allow programmatic access to it and add it to the ci-static-websites IAM group. Then take note of the access key id and the secret access key since you’ll need those later.

Adding deploy keys

To deploy websites we don’t use GitHub tokens (since they don’t have granular access scoping) but a deploy key with write access unique for each repository. To setup the deploy key you need to be an administrator on the repository, clone the simpleinfra repository and run this command:

$ cargo run --bin setup-deploy-keys rust-lang/repo-name

The command requires the GITHUB_TOKEN (you can generate one here) and the TRAVIS_TOKEN (you can see yours here) to be present. It will generate a brand new key, upload it to GitHub and configure Travis CI to use it if the repo is active there.

Configuring Travis CI

To actually deploy the website, this snippet needs to be added to your .travis.yml (please replace the contents of RUSTINFRA_DEPLOY_DIR and RUSTINFRA_CLOUDFRONT_DISTRIBUTION):

env:
  RUSTINFRA_DEPLOY_DIR: path/to/be/deployed
  RUSTINFRA_CLOUDFRONT_DISTRIBUTION: ABCDEFGHIJKLMN
import:
  - rust-lang/simpleinfra:travis-configs/static-websites.yml

You will also need to set the contents of the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables on the Travis CI web UI with the credentials of the IAM user you created earlier. The secret access key must be hidden from the build log, while the access key id should be publicly visible.

Configuring Azure Pipelines

To actually deploy the website, this snippet needs to be added at the top of your pipeline's YAML file:

resources:
  repositories:
    - repository: rustinfra
      type: github
      name: rust-lang/simpleinfra
      endpoint: rust-lang

Then you can add this steps when you want to execute the deploy (please replace the contents of deploy_dir and cloudfront_distribution):

- template: azure-configs/static-websites.yml@rustinfra
  parameters:
    deploy_dir: path/to/output
    # Optional, only needed if GitHub pages is behind CloudFront
    cloudfront_distribution: AAAAAAAAAAAAAA

You will also need to set the following environment variables in the pipeline:

• GITHUB_DEPLOY_KEY: value outputted when adding the deploy key earlier (secret)
• AWS_ACCESS_KEY_ID: access key ID of the IAM user allowed to invalidate CloudFront (public)
• AWS_SECRET_ACCESS_KEY: access key of the IAM user allowed to invalidate CloudFront (secret)

Infrastructure team documentation

This section contains the documentation about the services hosted and managed by the Rust Infrastructure Team. Most of the linked resources and instructions are only available to infra team members though.

AWS access for team members

Selected members of the Rust Team have access to the AWS account of the project. This includes both members of the Infrastructure Team and members of teams with services hosted on AWS.

This document explains how to access our AWS account, and how to interact with it. If you're a infrastructure team member and you need to setup or revoke access for another person, read the "AWS access management" page.

Setting up your user after receiving the credentials

The first thing you need to do after receiving your credentials is changing the password and enabling 2-factor authentication: until you do these things, access will be restricted automatically to just the permissions needed to configure 2FA.

Sign into the console with the temporary credentials given to you by the infrastructure team member who created the user. You'll be prompted to change the temporary password: change it and log in again. Then, go to the "My Security Credentials" page, located in the dropdown at the top:

Scroll down and click the "Assign MFA device" button. Choose "Virtual MFA device" (which is classic TOTP) and configure it with your authenticator app. Once you're done, log out of the console and log in again to gain access to the resources you're authorized to use.

Do not choose "U2F security key", even if you own one: due to limitations of the AWS API, that would prevent you from using the CLI, restricting your access to the console alone.

Using the AWS console

The AWS console provides a visual interface to most of the resources in our AWS account.

Using the AWS CLI

The AWS CLI allows you to interact with our AWS account from a terminal or a script. To set it up the first time, follow Amazon's documentation to install it and configure your credentials. The CLI doesn't use your console password to authenticate: you'll need to create an access key from the "My Security Credentials" page on the console.

2-factor authentication

To ensure the security of our AWS account, 2-factor authentication is required to interact with the CLI. The Infrastructure Team developed a script that eases the authentication process by creating a temporary session validated with 2FA for the current shell. The session expires in 12 hours, and it's valid for an unlimited number of invocations.

To use the script, clone the rust-lang/simpleinfra repository in a directory. Then, every time you need to use the AWS CLI run this command in your shell:

eval $(~/PATH/TO/SIMPLEINFRA/aws-creds.py)

That command will prompt you for your 2FA code, and it will set a few environment variables in the current shell with the temporary credentials. You'll need to run the command again after 12 hours, or if you want the credentials on another shell.

Plaintext credentials

By default, AWS CLI stores your credentials (including the secret key) in the ~/.aws/credentials file, without any kind of encryption. While the danger of having plaintext credentials stored in your home directory is partially mitigated by the 2FA requirement, it'd be best not to store them anyway.

If you use a password manager with a CLI interface, an approach you can take to avoid the problem is to store your credentials in the password manager, and configure the CLI to call your password manager to fetch the credentials when needed.

AWS access management

This document explains how to setup and manage AWS access for Rust team members. If you're a team member and you need to access AWS with your existing credentials, or you have received your credentials for the first time, check out the "AWS access for team members" page.

Granting access

To grant access to a person, go to team-members-access/_users.tf in the Terraform configuration and add the new user to it, specifying which teams they should be a member of. The user will be created as soon as you apply the configuration.

By default, there will be no credentials attached to the user. To allow the user to log in, go to the IAM console, open the security credentials page of the user you just created, and enable a console password. Let AWS generate a random one, and require the password to be changed on first login.

Finally communicate to the user that they can join with the generated password, and to follow the "AWS access for team members" page to learn how to enable 2FA and gain access to their account.

Revoking access

To revoke access from a person, log into the IAM console, open the security credentials page of the user you want to delete, and:

• Disable console access by clicking "Manage" on the console password
• Disable 2-factor authentication by clicking "Manage" on the assigned MFA device
• Remove all the access keys, including inactive ones, by clicking the "x".

Once all the access was removed from the console, go to team-members-access/_users.tf in the Terraform configuration, remove the user and apply the configuration.

Selection of AWS Regions

The Rust project has deployed a lot of resources on AWS, and most of them are in us-west-1. As we are growing our footprint and expand to more international locations, we are reconsidering which regions we want to use.

Please note that this is mainly for new resources that we are deploying, such as new AWS accounts. Existing resources might get migrated, but this is a significant effort that might not be worth it given our limited time.

Selection Criteria

We have two criteria that we use to make this decision:

• Price - Pricing differs between regions, and we can reduce our costs by deploying to cheaper regions.
• Location - We want to host our services close to most of our users. But given that Rust is used globally, we won't be able to satisfy everyone.

Price

Looking at the current distribution of our bill, outbound traffic is by far the most expensive item. This severely limits the price savings we might enjoy by switching to a cheaper region.

Even if we assume that we will be able to significantly reduce our outbound traffic cost on AWS (e.g. by moving to Fastly), the difference between regions is not massive.

Locations

Because most of our traffic comes from the US, we want to run most of our infrastructure here. The following regions are interesting to us:

• us-east-1 or us-east-2 (cheaper)
• us-west-1 (already in use)

Services we want to distribute more globally, e.g. the dev-desktops, we also want to deploy to Europe. Here, the following regions seem the most reasonable:

• eu-west-1 (cheaper)
• eu-central-1 (more central location)

Decision

We decided to use the following regions for new resources:

• us-east-2 - Given that most of our infrastructure is hosted in the US, we want to use a cheaper region here to benefit at least a little bit.
• eu-central-1 - Since we're not deploying that many resources to Europe, we want to optimize for location here.

When deploying new resources, they should be deployed to us-east-2 by default. Only resources that need to be geographically distributed should be deployed to eu-central-1.

Bastion server

• FQDN: bastion.infra.rust-lang.org
• Ansible playbook to deploy this server.
• Terraform configuration to create AWS resources.
• Instance metrics (only available to infra team members).

Logging into servers through the bastion

To improve the security of our infrastructure it's not possible to connect directly to a production server with SSH. Instead, all connections must come from a small server called the "bastion", which only allows connections from a few whitelisted networks and logs any connection attempt.

To log into a server through the bastion you can use SSH's -J flag:

ssh -J bastion.infra.rust-lang.org servername.infra.rust-lang.org

It's also possible to configure SSH to always jump through the bastion when connecting to a host. Add this snippet to your SSH configuration file (usually located in ~/.ssh/config):

Host servername.infra.rust-lang.org
    ProxyJump bastion.infra.rust-lang.org

Please remember the bastion server only allows connections from a small list of IP addresses. Infra team members with AWS access can change the whitelist, but it's good practice to either have your own bastion server or a static IP address.

The SSH keys authorized to log into each account are stored in the simpleinfra repository. Additionally, people with sensitive 1password access can use the master key stored in the vault to log into every account, provided their connection comes from any whitelisted IP.

Common maintenance procedures

Adding a new user to the bastion server

To add a new user to the bastion you need to add its key to a file named <username>.pub in ansible/roles/common/files/ssh-keys, and change the Ansible playbook adding the user to the list of unprivileged users. Please leave a comment clarifying which servers the user will have access to.

Once that's done apply the playbook and add a new whitelisted IP address.

Adding a whitelisted IP

Due to privacy reasons, all the static IP addresses of team members with access to the bastion are stored on AWS SSM Parameter Store instead of public git repositories. To add an IP address you can run this command (taking care of replacing USERNAME and IP_ADDRESS with the proper values):

aws ssm put-parameter --type String --name "/prod/bastion/allowed-ips/USERNAME" --value "IP_ADDRESS/32"

You'll also need to add the username to the list in terraform/bastion/firewall.tf (local variable allowed_users). Once you made all the needed changes you wanted you need to apply the Terraform configuration.

Updating a whitelisted IP

Due to privacy reasons, all the static IP addresses of team members with access to the bastion are stored on AWS SSM Parameter Store instead of public git repositories. To update an IP address you can run this command (taking care of replacing USERNAME and IP_ADDRESS with the proper values):

aws ssm put-parameter --overwrite --type String --name "/prod/bastion/allowed-ips/USERNAME" --value "IP_ADDRESS/32"

Once you made all the needed changes you wanted you need to apply the Terraform configuration.

Removing a whitelisted IP

Due to privacy reasons, all the static IP addresses of team members with access to the bastion are stored on AWS SSM Parameter Store instead of public git repositories. To remove an IP address you can run this command (taking care of replacing USERNAME with the proper value):

aws ssm delete-parameter --name "/prod/bastion/allowed-ips/USERNAME"

You'll also need to remove the username from the list in terraform/bastion/firewall.tf (local variable allowed_users). Once you made all the needed changes you wanted you need to apply the Terraform configuration.

Bors

The infrastructure team manages an instance of Homu called "Bors", to be used by repositories inside the rust-lang organization. The instance is available at bors.rust-lang.org, and is backed by the @bors GitHub account.

The service is configured with Terraform, and it's automatically deployed from the rust-lang/homu repository onto our ECS cluster.

Maintenance procedures

Fixing inconsistencies in the queue

Homu is quite buggy, and it might happen that the queue doesn't reflect the actual state in the repositories. This can be fixed by pressing the "Synchronize" button in the queue page. Note that the synchronization process itself is a bit buggy, and it might happen that PRs which were approved but failed are re-approved again on their own.

Adding a new repository to bors

There are multiple steps needed to add a repository to our Bors instance:

1. The @bors GitHub account needs to be granted write access to the repository.

2. Each CI provider needs to have a single GitHub Check Run to gate on. This is not provided by default on GitHub Actions, but it can be simulated with these two jobs, which will generate a bors build finished check:

   end-success:
     name: bors build finished
     if: success()
     runs-on: ubuntu-latest
     needs: [ALL, OTHER, JOBS]
     steps:
       - name: Mark the job as successful
         run: exit 0
   
   end-failure:
     name: bors build finished
     if: "!success()"
     runs-on: ubuntu-latest
     needs: [ALL, OTHER, JOBS]
     steps:
       - name: Mark the job as a failure
         run: exit 1
   

   Make sure to replace [ALL, OTHER, JOBS] with a list of all the jobs you want to gate on.

   These jobs need to run on specific branches (auto and try) so it's necessary to add those branches to the list of branches tested by the CI provider. For GitHub Actions that looks like this:

   on:
      push:
          branches: [ 
            auto,   # Added for bors
            try     # Added for bors
         ]
   

3. Add the repository name to the bors permissions array in the team repository, and grant the bors.REPOSITORY.review permission to the right teams or people. You can see an example of adding bors permissions to a team here.

4. Add the repository to the repositories map in the Terraform configuration file. This will create a webhook and inject its secret key in the bors execution environment.

5. Add the repository to the Bors configuration, taking inspiration from other repositories. Note that the environment variables used in that config will be set automatically as long as you completed step 3 above.

6. Give it a test by commenting @bors ping in any PR. If you get a response back, you can then try to approve the PR with @bors r+.

Content Delivery Networks

Users of the Rust programming language interact with the infrastructure of the project in various different ways. They access the project's website and documentation, query the crates index, and download Rust releases and crates. These resources are hosted by the Rust project and served through a Content Delivery Network (CDN).

This document outlines why we use CDNs, for what, and how we have set them up.

Objectives

We have three goals for our use of CDNs in our infrastructure:

1. Reduce costs of outbound traffic through cheaper pricing and caching
2. Reduce load on origin servers to save compute resources
3. Provide a way to rewrite legacy URLs for some resources

Reducing Costs

As an open source project, we have to be very mindful of our infrastructure costs. Outbound traffic is by far one of the most expensive items on our monthly bills, and one that will continue to increase as Rust gets more popular.

Cloud providers typically charge different rates for outbound traffic based on the service. For example, serving data straight from Amazon S3 is more expensive than serving the same data through an Amazon CloudFront distribution. This is why we now use a CDN by default, even for services that can't make use of other features of a CDN such as caching.

Infrastructure

Most of the project's resources are hosted on AWS. Static content is stored in Amazon S3, while dynamic content is loaded from a server. Both types of content are served through Amazon CloudFront, the Content Delivery Network of AWS.

When a user access a resource, e.g. they are trying to download a crate, they will access the resource through the CDN. Different distributions map domain names to a configuration and a backend (called the origin). For example, downloading a crate from static.crates.io goes through a distribution that fetches the crate from an S3 bucket and then caches it for future requests.

                             ┌──► S3 (static content)
                             │
User ───────► CloudFront ────┤
                             │
                             └──► Server (dynamic content)

Distributions

There are many distributions, all of which are configured in the rust-lang/simpleinfra repository. However, their usage is very unevenly distributed. The following distributions are the most important ones for the project, both in terms of traffic and criticality for the ecosystem.

Rust Releases

Whenever a user installs or updates Rust, pre-compiled binaries are downloaded from static.rust-lang.org. The same is true when Rust is installed in a CI/CD pipeline, which is why this distribution has by far the highest traffic volume.

Rust binaries are static and are stored in Amazon S3, from where they are served by the CloudFront distribution.

The distribution for static.rust-lang.org has a custom router that runs in a AWS Lambda function. The router provides a way to list files for a release and rewrites the legacy URL for rustup.sh.

The cache for Rust releases is invalidated nightly.

Crates

Similar to Rust releases, crates are served from as static content from static.crates.io. While still being the second-largest distribution in our infrastructure, it is much smaller than the releases.

Crates are static and stored in Amazon S3, and served through a CloudFront distribution.

Crater agents

• Source code: rust-lang/crater
• Hosted on:
  • crater-aws-1.infra.rust-lang.org (behind the bastion -- how to connect)
  • crater-azure-1.infra.rust-lang.org (behind the bastion -- how to connect)
• Maintainers: pietroalbini
• Application metrics (only available to infra team members).
• Instance metrics (only available to infra team members):
  • crater-aws-1.infra.rust-lang.org
  • crater-azure-1.infra.rust-lang.org

Service configuration

Crater agents are servers with our standard configuration running a Docker container hosting the agent. A timer checks for updates every 5 minutes, and if a newer Docker image is present the container will automatically be updated and restarted. This service is managed with Ansible.

Common maintenance procedures

Starting and stopping the agent

The agent is managed by the container-crater-agent.service systemd unit. That means it's possible to start, stop and restart it with the usual systemctl commands:

systemctl stop container-crater-agent.service
systemctl start container-crater-agent.service
systemctl restart container-crater-agent.service

Inspecting the logs of the agent

Logs of the agents are forwarded and collected by journald. To see them you can use journalctl:

journalctl -u container-crater-agent.service

Manually updating the container image

The container is updated automatically every 5 minutes (provided a newer image is present). If you need to update them sooner you can manually start the updater service by running this command:

systemctl start docker-images-update.service

Custom GitHub Actions runners

The Infrastructure Team manages a pool of self-hosted GitHub Actions runners, meant to be used by whitelisted repositories that need to run tests on platforms not supported by the GitHub-hosted runners. We're currently running the following machines:

• ci-arm-1.infra.rust-lang.org: AArch64 runners, hosted on packet (configuration).

The server configuration for the runners is managed with Ansible (playbook, role), and the source code for the tooling run on the server is in the gha-self-hosted repository.

Please get in touch with the Infrastructure Team if you need to run builds on this pool for your project in the rust-lang organization.

Maintenance procedures

Updating the GitHub Actions runner version

Our self-hosted CI runs on a custom fork of the GitHub Actions runner, which improves the security of the setup. The fork needs to be manually rebased every time a new version comes out though, and that needs to be done relatively quickly to prevent CI from stopping¹.

Once a new release of actions/runner is out, clone rust-lang/gha-runner and fetch the new tag pushed to the upstream repository. Then, rebase the changes on top of the latest tag:

git rebase --onto ${NEW_TAG} ${OLD_TAG} ${OLD_TAG}-rust${N}

For example, if the new tag is v2.275.0, the old tag is v2.274.2 and there were two releases of our fork, the command to execute would be:

git rebase --onto v2.275.0 v2.274.2 v2.274.2-rust2

The last commit to rebase will conflict, as that commit updates the version number and the release notes. Add the -rust1 suffix to the new version number and remove the description of the changes from the changelog (keeping the "Fork of the GitHub Actions runner used by the Rust Infrastructure Team." sentence). Once the rebase is complete force-push the commits to main.

After you force-push the new commits to main you're done! CI will create a tag, build the release, upload it to GitHub Releases, and automatically push a commit to rust-lang/gha-self-hosted bumping the pinned runner version to download in the images. The servers will then shortly pull the latest changes, rebuild the images and restart idle VMs.

Changing the instances configuration

The set of instances available in each host is configured through the Ansible configuration located in the simpleinfra repo:

ansible/envs/prod/host_vars/{hostname}.yml

You'll be able to add, remove and resize instances by changing that file and applying the changes:

ansible/apply prod gha-self-hosted

Forcing an update of the source code

The server checks for source code updates every 15 minutes, but it's possible to start such check in advance. You need to log into the machine you want to act on, and run the following command:

sudo systemctl start gha-self-hosted-update

If the contents of the images/ directory were changed, an image rebuild will also be started. The new image will be used by each VM after they finish processing the current job.

Forcing a rebuild of the images

The server automatically rebuilds the images every week, but it's possible to rebuild them in advance. You need to log into the machine you want to act on, and run the following command:

sudo systemctl start gha-self-hosted-rebuild-image

Managing the lifecycle of virtual machines

Each virtual machine is assigned a name and its own systemd unit, called gha-vm-{name}.service. For example, the arm-1-1 VM is managed by the gha-vm-arm-1-1.service systemd unit. You can stop, start and restart the virtual machine by stopping, starting and restarting the systemd unit.

Virtual machines are configured to restart after each build finishes.

Logging into the virtual machines

It's possible to log into the virtual machines from localhost to debug builds. This should be used as the last resort. Each VM binds SSH on a custom port on the host (configured in the host Ansible configuration), and allows access to the manage user (with password password). For example, to log into the VM with port 2201 you can run:

ssh manage@localhost -p 2201

Note that the VM image regenerates its own host key every time it boots, so you'll likely get host key mismatch errors when connecting to a freshly booted VM.

Accessing the out-of-band console for Packet servers

In the event that a bare metal server hosted on Packet becomes unreachable but is still marked as online, it's possible to access the out-of-band console over the serial port to get a root shell.

To access it, retrieve the root password configured on the server with:

aws ssm get-parameter --name /prod/ansible/HOSTNAME/root-password --with-decryption --query 'Parameter.Value' --output text

For example, to get the root password of ci-arm-1, run:

aws ssm get-parameter --name /prod/ansible/ci-arm-1/root-password --with-decryption --query 'Parameter.Value' --output text

Then, log into the packet console, navigate to the server page and click the "out-of-band console" button at the top right: the SSH command to use will be shown. Once you run the command you will be asked to login on the server: use root as the username and the password you fetched earlier as the password.

To exit the out-of-band console, type a new line followed by ~..

Dev Desktops

The dev desktops provide maintainers and contributors to the Rust Project with free access to high-powered cloud compute. They are part of the Cloud Compute Program by the Rust Foundation.

How to apply to the program

At this time, access to the program and the compute instances is limited to maintainers and core contributors of the Rust Project. While the program is under development, it is invite-only.

If you feel like your work on the Rust project would be significantly improved by access to a powerful build machine, reach out to infra@rust-lang.org with the following information:

• Your GitHub handle
• A short description of how you would use and benefit from the dev desktops

How to connect to a dev desktop

Each user has their own account on the dev desktops. The account is named after the user’s GitHub handle, with gh- as a prefix. For example, a user with the GitHub handle user will have a user account with the name gh-user on the dev desktop.

Users can connect to the dev desktop with SSH. The dev desktops use public key authentication, and automatically fetch the user’s public keys from GitHub.

You can connect to the instance with the following command:

ssh <your-username>@<name>.infra.rust-lang.org

Replace <name> with the machine name from the table at the top of the page. For example, connect to dev-desktop-eu-1 using the hostname dev-desktop-eu-1.infra.rust-lang.org.

If you don’t have a public key on GitHub, read the following guides that explain how to create an SSH key and add it to your GitHub account. It might take a few minutes after the key has been added before the dev desktops get updated.

• Generating a new SSH key and adding it to the ssh-agent
• Adding a new SSH key to your GitHub account

How to set up your account

When connecting to the machine for the first time, there are a few things you might want to do.

First, check that your Git username and email are configured correctly.

git config -l --global

You can configure your username and email address with:

git config --global user.name "Your name"
git config --global user.email "your-email"

How to install a Rust toolchain

The dev desktops don’t have Rust pre-installed, but instead make it easy to install a specific toolchain from a local repository or worktree.

First, you want to run the following command to install rustup:

/usr/local/bin/init.sh

If you don't want or need to work with your own version of Rust, you can skip the next section and start working.

If you haven't done so yet, open the rust-lang/rust repository on GitHub and create a fork in your personal account. Then connect to the dev desktop and run the following script:

/usr/local/bin/setup_rust.sh

The script will clone your personal fork to the dev desktop, check out the latest version from rust-lang/rust, and compile it. Once that's done, it will link the stages so that you can work with them locally.

The directory contains more scripts to manage worktrees and Rust versions. Run help.sh to get a list and a short description of them.

How to interact with GitHub

The dev desktops are designed to work with repositories on GitHub that belong to your user account. A GitHub App is used to protect your credentials and give you granular control over the repositories that the dev desktops can access.

First, go to https://github.com/apps/rust-cloud-vms to give the app access to your repositories. It's recommended to only grant access to the repositories that you want to use on the dev desktop, e.g. your fork of rust-lang/rust.

Then connect to the dev desktop and clone the repository that you want to work on with HTTPS. From there, you can work with the repository like you would normally do.

Under the hood, the GitHub App acts as a credentials helper for Git and generates temporary access tokens that are scoped to the permissions that you have granted the application. If you get an error, review the permissions and ensure that the app is allowed to access your repository.

How to set up remote development in Visual Studio Code

Most modern code editors provide support for remote development via SSH. This can be used to write code locally, but execute it inside the dev desktop. While the configuration will differ slightly, the following example for Visual Studio Code should be applicable to other editors as well.

Setting up remote development with VS Code is pretty straightforward, and is described in detail in VS Code’s documentation: Remote Development using SSH. In summary:

1. SSH into the dev desktop and clone the repository that you want to work on to a local folder
2. Then open VS Code on your machine and install the Remote Development Extension Pack
3. Open the command palette and search for “Remote-SSH: Connect to host”
4. Enter your username and the instance name (<your-username>@<instance>)
5. Select the path for the cloned repository from step 1
6. Install any extensions that you want to run on the server (e.g. rust-analyzer)
7. Use VS Code to run or debug the code remotely

How to give feedback and report issues

If you experience any problems with the dev desktops, or have feedback and suggestions, get in touch with the infrastructure team:

#t-infra on Zulip

We might ask you to create an issue in the rust-lang/simpleinfra repository.

Github App for pushing to github from the dev-desktops

These instructions are for server-side setup and debugging of the dev-desktop github app. The user only needs to be directed to the app installation URL and everything should just work for them.

We're using the python github library for all github operations. You can find the docs at https://pygithub.readthedocs.io/en/latest/introduction.html

How to setup an App

1. Go to https://github.com/settings/apps
2. New Github App
3. Fill out metadata (name and url)
4. disable WebHook checkbox
5. Set Contents - Repository contents, commits, branches, downloads, releases, and merges. to read/write
6. Set Workflows - Update GitHub Action workflow files. to read/write
7. Set to "enable on any account"
8. Create App
9. Go to https://github.com/settings/apps/{your_app_name_here} and copy the App ID into app_id.txt (same folder as gen_temp_access_token.py)

How to generate a .pem file for your App

1. Go to https://github.com/settings/apps/{your_app_name_here}#private-key and generate a private key
2. Download starts, save it to somewhere private.
3. copy the .pem file into the same folder as the gen_temp_access_token.py and name it dev-desktop.private-key.pem

How to install the app for a user

1. direct the user to https://github.com/settings/apps/{your_app_name_here}/installations
2. let them install it on the org/user they want to and restrict to the repositories they want to use

How to generate a temporary access token for a specific user

1. invoke gen_temp_access_token.py <github_username> <github_repository_name>

Integration into git command line

We're using credential-helpers. For debugging a credential helper, have it in userspace and invoke it with

git -c credential.helper -c credential.UseHttpPath=true /path/to/helper push origin branch

Note that this does not work for remotes that are registered with ssh urls. You must use https!

The first command line argument is get, store or remove. In our case, we just abort (exit(0)) for everything but get, as we regenerate credentials on every invocation anyway.

The actual arguments are passed via stdin and usually look like

protocol=https
host=github.com
path=your_repo.git

Discord moderation bot

• Source code: rust-lang/discord-mods-bot
• Hosted on: rust-ecs-prod ECS cluster
• Maintainers: technetos

The bot is hosted on the rust-ecs-prod ECS cluster, on the project's AWS account, with the discord-mods-bot service name. Its container image is stored in a ECR repository with the same name, and its data is stored in the shared RDS PostgreSQL instance.

Automatic deploys are setup from the rust-lang/discord-mods-bot GitHub repository.

The Discord bot account is rustbot#4299. pietroalbini, Mark-Simulacrum, alexcrichton and aidanhs have access to the developer portal.

Common maintenance procedures

Instructions on how to manage ECS services are available here.

Domain names and DNS

All the DNS records of the domains owned by the Rust Infrastructure team are hosted on AWS Route 53, and can be tweaked by members of the team. This document contains instructions for them on how to make changes.

• Changing DNS records of a domain managed with Terraform
• Managing DNS for a new domain with Terraform
• Adding subdomain redirects
• Transferring domain names to Rust

Changing DNS records of a domain managed with Terraform

Warning: not all domain names are yet managed with Terraform. In the console, if a zone's comment doesn't start with [terraform] you'll need to make changes manually from the UI. Work is underway to migrate every domain to Terraform though.

Warning: terraform/services/dns only contains the definition of DNS records pointing to resources managed outside of Terraform. When Terraform manages a resource it will automatically add the required records on its own. See the service's documentation to learn where its Terraform configuration lives.

DNS records are managed in the terraform/services/dns directory of our Terraform configuration. A file named after the domain name, ending in .tf, exists for each managed domain, and it contains some basic information plus its records.

The configuration supports adding A, CNAME, MX and TXT records. Inside the module definition contained in the domain's file, each record type has its own map: the map keys are the names of the records, while the values are a list of record values.

For example, to add a pages.rust-lang.org CNAME pointing to rust-lang.github.io you'll need to add this to terraform/services/dns/rust-lang.org:

module "rust_lang_org" {
  # ...

  CNAME = {
    "pages.rust-lang.org." = ["rust-lang.github.io"],
    # ...
  }
}

Once you made all the changes you can apply them with:

terraform apply

Managing DNS for a new domain with Terraform

Setting up Terraform to manage the DNS records of a new domain name involves a few steps. First of all you need to decide the identifier used inside Terraform for that domain. By convention, the identifier is the domain name itself with . and - replaced with _. For example rust-lang.org becomes rust_lang_org.

Then you can create a file in terraform/services/dns named after the domain name, ending in .tf, with this content (take care of replacing the placeholders):

module "<IDENTIFIER>" {
  source = "./domain"

  domain = "<DOMAIN-NAME>"
  comment = "<COMMENT-FOR-THE-DOMAIN>"
  ttl = 300
}

Finally you need to output the ID of the Route53 zone, allowing other parts of our Terraform configuration to add records. Add this snippet to terraform/services/dns/outputs.tf:

# ...

output "zone_<IDENTIFIER>" {
  value = module.<IDENTIFIER>.zone_id
}

Once you're done you can apply the changes with:

terraform init
terraform apply

Adding subdomain redirects

Our Terraform configuration supports creating redirects from an arbitrary number of subdomains we control to an URL. Redirects are created with these pieces of infrastructure:

• A S3 bucket for each set of redirects, named rust-http-redirect-<HASH>. The bucket has website hosting enabled, configured to redirect all the incoming requests to the chosen URL. This allows implementing redirects without an underlying server.

• An ACM certificate (plus the DNS records to validate it) for each set of redirects, with all the sources as alternate names. This is used to enable HTTPS redirects.

• A CloudFront distribution for each set of redirects to support HTTPS requests, using the previously generated ACM certificate and forwarding requests to the S3 bucket.

• Route53 records for each redirect in the related zones: CNAMEs for subdomains, and ALIASes for apex domains.

All the redirects are defined in terraform/redirects.tf, with a module for each destination URL. Either create a new module if you need to redirect to a new URL, or add a new subdomain to an existing module. See an example module here (take care of replacing the placeholders):

module "redirect_<IDENTIFIER>" {
  source = "./modules/subdomain-redirect"
  providers = {
    aws       = "aws"
    aws.east1 = "aws.east1"
  }

  to = "<DESTINATION-URL>"
  from = {
    "<SUBDOMAIN-1>" = module.dns.zone_<DOMAIN-1-IDENTIFIER>,
    "<SUBDOMAIN-2>" = module.dns.zone_<DOMAIN-2-IDENTIFIER>,
  }
}

Once you made all the changes you can apply the configuration with:

terraform init
terraform apply

Note that each change is going to take around 15 minutes to deploy, as CloudFront distribution changes are really slow to propagate. Also, it's normal to see a bunch of resources being recreated when a domain is added or removed from an existing redirect, as the ACM certificate will need to be regenerated.

Transferring domain names to Rust

These are the steps a member of the infrastructure team needs to take to transfer a domain name to the Rust project's registrar:

1. Ask inside the infrastructure team if this is a domain name the project wants to own. In some more complicated cases this will need to be escalated to the core team.

2. If the domain name doesn’t already use AWS Route 53 as its nameserver, ask the current owner of the domain a list of all the DNS records that will need to be migrated. Then, add all the records to a new hosted zone on Route 53 before the transfer of the domain. See the section below on transferring DNS for more information on this step.

3. Ask the current owner to unlock the domain name for transfer, and get the transfer code from them. The transfer code is key to transferring the domain, so avoid receiving it on public communication platforms.

4. Go to the Transfer Domain section of AWS Route 53 and enter the domain name. If it doesn’t give an error (which should detail which steps are missing) enter the transfer code you received earlier, and choose to use an existing Route 53 hosted zone (it should auto-complete the right one). Until the Rust Foundation is up, use Pietro’s details as the domain contacts. Finally review everything and complete the transfer process.

5. Tell the current owner to wait for an email from their registrar, which will ask to click on a link to confirm the domain name transfer.

6. The transfer process will take a while. Once admin@rust-lang.org receives an email telling the domain has been transferred you’re done! 🎉🎉🎉

Transferring DNS

Most domain names use their registrar as the DNS server, but that means that once the domain is transferred away the old registrar also stops serving DNS traffic. Because of that we need to ensure all the DNS records are correctly copied over to AWS Route 53 before actually starting the transfer process.

Explicitly ask the current domain owner for all the A, AAAA, CNAME, TXT and MX records. Everything except the MX records needs to be copied to the Terraform DNS configuration (create a new file for the domain name, and take inspiration from the other domain names).

If you notice some of the records are referring to HTTP redirect services provided by the current registrar then those will have to wait until the domain has been transferred. Once the transfer occured, add a new domain redirect on Terraform. This has to be done after the transfer to be able to request the TLS certificate for the HTTPS redirect.

If the domain has MX records those will need to be migrated to Mailgun. Go to Mailgun and add the domain name there. Ensure it’s in the US region, it uses shared IPs, and it has a 1024 bit DKIM key (the 2048 keys do not fit into a single AWS Route 53 record). Then copy all the records except the CNAME tracking one over to the Terraform DNS configuration, and wait for the domain to be transferred. Once the transfer happens go back to Mailgun and verify the DNS settings for the domain. Finally, add the domain to the team repository’s config.toml and create the mailing lists you need through the usual process.

docs.rs

• Source code: rust-lang/docs.rs
• Hosted on: docsrs.infra.rust-lang.org (behind the bastion -- how to connect)
• Maintainers: Joshua Nelson, Pietro Albini
• Instance metrics (only available to infra team members).
• Application metrics (only available to infra team members).
• Common maintenance procedures

ECS services management

Some applications running on the project's infrastructure are hosted in ECS clusters on our AWS account. This document explains the common maintenance procedures one should follow when operating them. Most of the actions explained here require AWS access.

Note: our ECS cluster is located in the Northern California (us-west-1) AWS region. Make sure it's the selected region when interacting with the AWS console.

Inspecting the logs

Logs for applications hosted on ECS are stored in CloudWatch Logs, and can be inspected in the AWS Console. Open the console, go to CloudWatch Logs and select the log group called /ecs/<service-name>. There are two ways to inspect the logs:

• If you need to look at the application as a whole, you can get an aggregated view by clicking the "View all log events" button (or, on the classic interface, "Search Log Group").

• If you need to debug a specific instance of a container, separate log streams for each running task are available. The streams are named after the container name and the task ID.

Logs are periodically purged (retention varies based on the specific application).

Restarting an application

To restart an application, you can force a new deployment without actually pushing any new code beforehand. To do so, run this command:

aws ecs update-service --cluster rust-ecs-prod --service <service-name> --force-new-deployment

Rolling back a deployment

To rollback a bad deployment you can run the aws-rollback.py script (stored in the simpleinfra repository) with your AWS credentials present in the shell. The script requires the name of the ECR container image repository as its first and only argument:

./aws-rollback.py <image-repository-name>

The script will show the list of images available in the repository, and asks for the image number to rollback to. Once that's inserted the script will point the latest tag to the image you chose, and if an ECS service with the same name as the repository exists that service will be restarted too.

Deploying application changes

Each application stores its own Docker container in a ECR repository in our AWS account. You can deploy changes both manually and automatically (with GitHub Actions).

For production applications it's recommended to setup automatic deployment.

Manual deployments

To manually deploy a local build you first need it to tag your built image with its ECR name:

docker tag <image-tag> 890664054962.dkr.ecr.us-west-1.amazonaws.com/<repository-name>:latest

Then you can authenticate with ECR and push it:

$(aws ecr get-login --no-include-email --region us-west-1)
docker push 890664054962.dkr.ecr.us-west-1.amazonaws.com/<repository-name>:latest

Finally, you need to force a new deployment of the ECS service with:

aws ecs update-service --cluster rust-ecs-prod --service <service-name> --force-new-deployment

Automatic deployments with GitHub Actions

The infrastructure team prepared an action for GitHub Actions that automates deployments from CI. To use it, ask a team member to setup AWS credentials in your repository, and then add this snippet to your workflow:

- name: Build the Docker image
  run: docker build -t deploy-image .

- name: Deploy to production
  uses: rust-lang/simpleinfra/github-actions/upload-docker-image@master
  with:
    image: deploy-image
    repository: <ecr-repository-name>
    region: us-west-1
    redeploy_ecs_cluster: rust-ecs-prod
    redeploy_ecs_service: <service-name>
    aws_access_key_id: "${{ secrets.AWS_ACCESS_KEY_ID }}"
    aws_secret_access_key: "${{ secrets.AWS_SECRET_ACCESS_KEY }}"
  if: github.ref == 'refs/heads/<deploy-branch>'

Be sure to replace <ecr-repository-name>, <service-name> and <deploy-branch> with the correct values for your workflow. Once the workflow changes are merged in the branch you chose for deploys, any future commits pushed there will be deployed to the ECS cluster.

Monitoring

• Hosted on: monitoring.infra.rust-lang.org (behind the bastion -- how to connect)
• Maintainers: pietroalbini, infra team
• Public URL: grafana.rust-lang.org
• Ansible playbook to deploy this server.
• Instance metrics (only available to infra team members).

Service configuration

Our monitoring service is composed of three parts: Prometheus to scrape, collect and monitor metrics, Alertmanager to dispatch the alerts generated by Prometheus, and Grafana to display the metrics. All the parts are configured through Ansible.

The metrics are not backed up, as Prometheus purges them after 7 days anyway, but the Grafana dashboards are stored in a PostgreSQL database, which is backed up with restic in the rust-backups bucket (monitoring subdirectory). The password to decrypt the backups is in 1password.

Common maintenance procedures

Scrape a new metrics source

Prometheus works by periodically scraping a list of HTTP endpoints for metrics, written in its custom format. In our configuration the list is located in the prometheus_scrape section of the ansible/playbooks/monitoring.yml file in the simpleinfra repository.

To add a new metrics source, add your endpoint to an existing job or, if the metrics you're scraping are not related to any other job, a new one. The endpoint must be reachable from the monitoring instance. You can read the Prometheus documentation to find all the available options.

Create a new alert

Alerts are generated by Prometheus every time a custom rule defined in its configuration evaluates to true. In our configuration the list of rules is located in the prometheus_rule_groups section of the ansible/playbooks/monitoring.yml file in the simpleinfra repository.

To add a new alert you need to create an alerting rule either in an existing group or a new one. The full list of options is available in the Prometheus documentation.

Add permissions to a user

There are two steps needed to grant access to our Grafana instance to an user.

First of all, to enable the user to log into the instance with their GitHub account they need to be a member of a team authorized to log in. The list of teams is defined in the grafana_github_teams section of the ansible/playbooks/monitoring.yml file in the simpleinfra repository, and it contains a list of GitHub team IDs. To fetch an ID you can run this command:

curl -H "Authorization: token $GITHUB_TOKEN" https://api.github.com/orgs/<ORG>/teams/<NAME> | jq .id

Once the user is a member of a team authorized to log in they will automatically be added to the main Grafana organization with "viewer" permissions. For infrastructure team members that needs to be changed to "admin" (in the "Configuration" -> "Users"), otherwise leave it as viewer.

By default a viewer only has access to the unrestricted dashboards. To grant access to other dashboards you'll need to add them to a team (in the "Configuration" -> "Teams" page). It's also possible to grant admin privileges to the whole Grafana instance in the "Server Admin" -> "Users" -> "<username>" page. Do not grant those permissions except to trusted infra team members.

Additional resources

• Prometheus documentation
• Grafana documentation

rust-bots

• FQDN: bots.infra.rust-lang.org (behind the bastion -- how to connect)
• Instance metrics (only available to infra team members).

Common maintenance procedures

Adding a new domain

First, edit sudo vim /etc/nginx/nginx.conf to edit the nginx configuration to add the domain.

server {
    listen 443 ssl;
    listen [::]:443 ssl;
    server_name <domain>.infra.rust-lang.org; # Edit <domain> to match here

    location /.well-known/acme-challenge {
        root /home/ssl-renew/challenges;
    }

    location / {
        # configure the domain here
    }
}

Then run sudo -i -u ssl-renew vim renew.sh. Add a --domains line to the script with the domain you're adding.

Then, run the script: sudo -i -u ssl-renew ./renew.sh

How the Rust CI works

Rust CI ensures that the master branch of rust-lang/rust is always in a valid state.

A developer submitting a pull request to rust-lang/rust, experiences the following:

• A small subset of tests and checks are run on each commit to catch common errors.
• When the PR is ready and approved, the "bors" tool enqueues a full CI run.
• The full run either queues the specific PR or the PR is "rolled up" with other changes.
• Eventually a CI run containing the changes from the PR is performed and either passes or fails with an error the developer must address.

Which jobs we run

The rust-lang/rust repository uses GitHub Actions to test all the platforms we support. We currently have two kinds of jobs running for each commit we want to merge to master:

• Dist jobs build a full release of the compiler for that platform, including all the tools we ship through rustup; Those builds are then uploaded to the rust-lang-ci2 S3 bucket and are available to be locally installed with the rustup-toolchain-install-master tool; The same builds are also used for actual releases: our release process basically consists of copying those artifacts from rust-lang-ci2 to the production endpoint and signing them.
• Non-dist jobs run our full test suite on the platform, and the test suite of all the tools we ship through rustup; The amount of stuff we test depends on the platform (for example some tests are run only on Tier 1 platforms), and some quicker platforms are grouped together on the same builder to avoid wasting CI resources.

All the builds except those on macOS and Windows are executed inside that platform’s custom Docker container. This has a lot of advantages for us:

• The build environment is consistent regardless of the changes of the underlying image (switching from the trusty image to xenial was painless for us).
• We can use ancient build environments to ensure maximum binary compatibility, for example using older CentOS releases on our Linux builders.
• We can avoid reinstalling tools (like QEMU or the Android emulator) every time thanks to Docker image caching.
• Users can run the same tests in the same environment locally by just running src/ci/docker/run.sh image-name, which is awesome to debug failures.

The docker images prefixed with dist- are used for building artifacts while those without that prefix run tests and checks.

We also run tests for less common architectures (mainly Tier 2 and Tier 3 platforms) in CI. Since those platforms are not x86 we either run everything inside QEMU or just cross-compile if we don’t want to run the tests for that platform.

These builders are running on a special pool of builders set up and maintained for us by GitHub.

Almost all build steps shell out to separate scripts. This keeps the CI fairly platform independent (i.e., we are not overly reliant on GitHub Actions). GitHub Actions is only relied on for bootstrapping the CI process and for orchestrating the scripts that drive the process.

Merging PRs serially with bors

CI services usually test the last commit of a branch merged with the last commit in master, and while that’s great to check if the feature works in isolation it doesn’t provide any guarantee the code is going to work once it’s merged. Breakages like these usually happen when another, incompatible PR is merged after the build happened.

To ensure a master that works all the time we forbid manual merges: instead all PRs have to be approved through our bot, bors (the software behind it is called homu). All the approved PRs are put in a queue (sorted by priority and creation date) and are automatically tested one at the time. If all the builders are green the PR is merged, otherwise the failure is recorded and the PR will have to be re-approved again.

Bors doesn’t interact with CI services directly, but it works by pushing the merge commit it wants to test to a branch called auto, and detecting the outcome of the build by listening for either Commit Statuses or Check Runs. Since the merge commit is based on the latest master and only one can be tested at the same time, when the results are green master is fast-forwarded to that merge commit.

The auto branch and other branches used by bors live on a fork of rust-lang/rust: rust-lang-ci/rust. This was originally done due to some security limitations in GitHub Actions. These limitations have been addressed, but we've not yet done the work of removing the use of the fork.

Unfortunately testing a single PR at the time, combined with our long CI (~3 hours for a full run)¹, means we can’t merge too many PRs in a single day, and a single failure greatly impacts our throughput for the day. The maximum number of PRs we can merge in a day is around 8.

The large CI run times and requirement for a large builder pool is largely due to the fact that full release artifacts are built in the dist- builders. This is worth it because these release artifacts:

• allow perf testing even at a later date
• allow bisection when bugs are discovered later
• ensure release quality since if we're always releasing, we can catch problems early

Bors runs on ecs and uses a sqlite database running in a volume as storage.

Rollups

Some PRs don’t need the full test suite to be executed: trivial changes like typo fixes or README improvements shouldn’t break the build, and testing every single one of them for 2 to 3 hours is a big waste of time. To solve this we do a "rollup", a PR where we merge all the trivial PRs so they can be tested together. Rollups are created manually by a team member using the "create a rollup" button on the bors queue. The team member uses their judgment to decide if a PR is risky or not, and are the best tool we have at the moment to keep the queue in a manageable state.

Try builds

Sometimes we need a working compiler build before approving a PR, usually for benchmarking or checking the impact of the PR across the ecosystem. Bors supports creating them by pushing the merge commit on a separate branch (try), and they basically work the same as normal builds, without the actual merge at the end. Any number of try builds can happen at the same time, even if there is a normal PR in progress.

You can see the CI configuration for try builds here.

Which branches we test

Our builders are defined in src/ci/github-actions/ci.yml.

PR builds

All the commits pushed in a PR run a limited set of tests: a job containing a bunch of lints plus a cross-compile check build to Windows mingw (without producing any artifacts) and the x86_64-gnu-llvm-## non-dist builder (where ## is the system LLVM version we are currently testing). Those two builders are enough to catch most of the common errors introduced in a PR, but they don’t cover other platforms at all. Unfortunately it would take too many resources to run the full test suite for each commit on every PR.

Additionally, if the PR changes certain tools (or certain platform-specific parts of std to check for miri breakage), the x86_64-gnu-tools non-dist builder is run.

The try branch

On the main rust repo, try builds produce just a Linux toolchain using the dist-x86_64-linux image.

The auto branch

This branch is used by bors to run all the tests on a PR before merging it, so all the builders are enabled for it. bors will repeatedly force-push on it (every time a new commit is tested).

The master branch

Since all the commits to master are fast-forwarded from the auto branch (if they pass all the tests there) we don’t need to build or test anything. A quick job is executed on each push to update toolstate (see the toolstate description below).

Other branches

Other branches are just disabled and don’t run any kind of builds, since all the in-progress branches will eventually be tested in a PR.

Caching

The main rust repository doesn’t use the native GitHub Actions caching tools. All our caching is uploaded to an S3 bucket we control (rust-lang-ci-sccache2), and it’s used mainly for two things:

Docker images caching

The Docker images we use to run most of the Linux-based builders take a long time to fully build. To speed up the build, we cache the exported images on the S3 bucket (with docker save/docker load).

Since we test multiple, diverged branches (master, beta and stable) we can’t rely on a single cache for the images, otherwise builds on a branch would override the cache for the others. Instead we store the images identifying them with a custom hash, made from the host’s Docker version and the contents of all the Dockerfiles and related scripts.

LLVM caching with sccache

We build some C/C++ stuff during the build and we rely on sccache to cache intermediate LLVM artifacts. Sccache is a distributed ccache developed by Mozilla, and it can use an object storage bucket as the storage backend, like we do with our S3 bucket.

Custom tooling around CI

During the years we developed some custom tooling to improve our CI experience.

Rust Log Analyzer to show the error message in PRs

The build logs for rust-lang/rust are huge, and it’s not practical to find what caused the build to fail by looking at the logs. To improve the developers’ experience we developed a bot called Rust Log Analyzer (RLA) that receives the build logs on failure and extracts the error message automatically, posting it on the PR.

The bot is not hardcoded to look for error strings, but was trained with a bunch of build failures to recognize which lines are common between builds and which are not. While the generated snippets can be weird sometimes, the bot is pretty good at identifying the relevant lines even if it’s an error we've never seen before.

Toolstate to support allowed failures

The rust-lang/rust repo doesn’t only test the compiler on its CI, but also a variety of tools and documentation. Some documentation is pulled in via git submodules. If we blocked merging rustc PRs on the documentation being fixed, we would be stuck in a chicken-and-egg problem, because the documentation's CI would not pass since updating it would need the not-yet-merged version of rustc to test against (and we usually require CI to be passing).

To avoid the problem, submodules are allowed to fail, and their status is recorded in rust-toolstate. When a submodule breaks, a bot automatically pings the maintainers so they know about the breakage, and it records the failure on the toolstate repository. The release process will then ignore broken tools on nightly, removing them from the shipped nightlies.

While tool failures are allowed most of the time, they’re automatically forbidden a week before a release: we don’t care if tools are broken on nightly but they must work on beta and stable, so they also need to work on nightly a few days before we promote nightly to beta.

More information is available in the toolstate documentation.

GitHub Actions Templating

GitHub Actions does not natively support templating which can cause configurations to be large and difficult to change. We use YAML anchors for templating and a custom tool, expand-yaml-anchors, to expand the template into the CI configuration that GitHub uses.

This templating language is fairly straightforward:

• & indicates a template section
• * expands the indicated template in place
• << merges yaml dictionaries

Sentry

The infrastructure team manages a Sentry organization on sentry.io for the Rust Team to use. The instance is generously sponsored by Sentry, and this document explains how to use it.

Log into the instance

Every member of the rust-lang GitHub organization can authenticate in our Sentry instance, using their GitHub credentials. Visit the authentication page, click the "Single Sign-On" tab and enter the rust-lang Organization ID. You'll be then prompted to log with your GitHub Account!

If this is the first time signing into our Sentry organization, you might have to request access to the teams you're on. Once you request access, a member of the infrastructure team will approve it.

Request a new project

If you're a member of a Rust Team and you want to use Sentry for a project your team manages, you need to follow these steps:

1. If the project is public facing (i.e. people outside the team are supposed to access it) you need to contact the Core team to request support in amending the privacy policy, adding a note that your service is using Sentry too similar to the existing ones.

2. Once the privacy policy is sorted out (whenever needed), you can contact the infrastructure team to create a new project in the Sentry interface and potentially a new Sentry team.

3. Finally, you can integrate the Sentry SDK with your project.

Creating a new project

This section documents how the infrastructure team can actually create new projects when requested. You need to either have a personal Sentry account with "Owner" permissions, or access to the Sensitive 1Password vault (where the admin credentials are stored).

To create a project, authenticate in Sentry and visit the create new project page. Pick the technology stack the team is using, a relevant name and the team responsible for it (you can create new teams by clicking the "+" icon). Finally, if you created a new team, add the relevant people to it.

Language

This section documents meta processes by the language team.

External Links

• The language team has communications channels on Discord as well as Zulip.

RFC Merge Procedure

Once an RFC has been accepted (i.e., the final comment period is complete, and no major issues were raised), it must be merged. Right now this is a manual process, though just about anyone can do it (if you're not a subteam member, though, you'll have to open a PR rather than merge the RFC manually). Here is the complete set of steps to merge an RFC -- in some cases, not all the steps will be applicable.

Step 1: Open tracking issue

Open a tracking issue over on rust-lang/rust. Here is a template for the issue text. You'll have to adjust the various places labeled XXX with some suitable content (e.g., the name of the RFC, or the most appropriate team).

This is a tracking issue for the RFC "XXX" (rust-lang/rfcs#NNN).

**Steps:**

- [ ] Implement the RFC (cc @rust-lang/XXX -- can anyone write up mentoring
      instructions?)
- [ ] Adjust documentation ([see instructions on rustc-dev-guide][doc-guide])
- [ ] Stabilization PR ([see instructions on rustc-dev-guide][stabilization-guide])

[stabilization-guide]: https://rustc-dev-guide.rust-lang.org/stabilization_guide.html#stabilization-pr
[doc-guide]: https://rustc-dev-guide.rust-lang.org/stabilization_guide.html#documentation-prs

**Unresolved questions:**

XXX --- list all the "unresolved questions" found in the RFC to ensure they are
not forgotten

Add the following labels to the issue:

• B-rfc-approved
• C-tracking-issue
• the approriate T-XXX label

(If you don't have permissions to do so, leave a note cc'ing the appropriate team and asking them to do so.)

Step 2: Merge the RFC PR itself

In your local git checkout:

• Merge the RFC PR into master in your fork
• Add a commit that moves the file name from 0000- to its RFC number
• Edit the new file to include links to the RFC PR and the tracking issue you just created in the header
• Open a PR or push directly to the master branch on rust-lang/rfcs, as appropriate

Step 3: Leave a comment

Leave a final comment on the PR directing everyone to the tracking issue. Something like this, but feel free to add your own personal flavor (and change the team):

**Huzzah!** The @rust-lang/lang team has decided **to accept** this RFC.

To track further discussion, subscribe to the tracking issue here:
rust-lang/rust#41517

Step 4: Update the rendered link

Update the rendered link in the first post of the PR to the permanent home under https://github.com/rust-lang/rfcs/blob/master/text/.

(This way future visitors can open it easily after the PR branch is deleted.)

That's it, you're done!

Triage meeting procedure

This page documents how to run a lang team triage meeting, should you have the misfortune of being forced to do so.

Attending a meeting

If you would just like to attend a lang-team triage meeting, all you have to do is join the zoom call (the URL is attached to the calendar invite below).

Scheduling

Note that the scheduling for all meetings is recorded in the team calendar, links to which can be found on the rust-lang/lang-team repository.

Pre-triage

To start, we have a pre-triage meeting which occurs before the main meeting. This is not recorded. It is boring.

To execute this meeting you:

• Open the Current Meeting dropbox paper document
• Skim down the action items and look to see if there are any you know have been handled
  • they can be checked off and removed
• Skip down to the Triage section
• For each Triage section, click on the link and populate it with what you find
  • typically it is best to copy-and-paste the title of the issue, so that links remain intact
• For each item, click in and try to add a few notes as to the main topic
  • look for things where there isn't much discussion needed, or just reminders
  • these can be handled quickly in the meeting, or perhaps not at all
  • items that require more discussion will need time alotted for them

Main meeting

• Ping the team on discord @lang-team
• Begin the recording on Zoom, if you have acccess
  • If nobody has access to the recording, oh well, we don't do it every week
• Discuss item by item and take some notes on what was said
  • Add specific actions to the action items section above
  • If a consensus arises, make sure to create an action item to document it!
  • The goal should be that we leave some comment on every issue

After meeting

• Export the meeting file to markdown
  • you will need to cleanup "check boxes" -- Niko usually searches and replaces ^(\s*)[ ] with \1* [ ] or something like that to insert a * before them, which makes them valid markdown
• Upload video to youtube if applicable and get the URL
• Add the file to the minutes directory of rust-lang/lang-team repository with a file name like YYYY-MM-DD.md

Libs

This section documents meta processes by the Libs team.

Where to find us

The rust-lang/libs-team GitHub repository is the home of the Libs team. It has details on current project groups, upcoming meetings, and the status of tracking issues.

The Libs team hangs out primarily in the rust-lang Zulip these days in the #t-libs stream.

You can also find out more details about Zulip and how the Rust community uses it.

Maintaining the standard library

Everything I wish I knew before somebody gave me r+

This document is an effort to capture some of the context needed to develop and maintain the Rust standard library. It’s goal is to help members of the Libs team share the process and experience they bring to working on the standard library so other members can benefit. It’ll probably accumulate a lot of trivia that might also be interesting to members of the wider Rust community.

This document doesn't attempt to discuss best practices or good style. For that, see the API Guidelines.

Contributing

If you spot anything that is outdated, under specified, missing, or just plain incorrect then feel free to open up a PR on the rust-lang/rust-forge repository!

Terms

• Libs. That's us! The team responsible for development and maintenance of the standard library (among other things).
• Pull request (PR). A regular GitHub pull request against rust-lang/rust.
• Request for Comment (RFC). A formal document created in rust-lang/rfcs that introduces new features.
• Tracking Issue. A regular issue on GitHub that’s tagged with C-tracking-issue.
• Final Comment Period (FCP). Coordinated by rfcbot that gives relevant teams a chance to review RFCs and PRs.

If you’re ever unsure…

Maintaining the standard library can feel like a daunting responsibility! Through automated reviewer assignment via triagebot, you’ll find yourself dropped into a lot of new contexts.

Ping the @rust-lang/libs team on GitHub anytime. We’re all here to help!

If you don’t think you’re the best person to review a PR then use triagebot to assign it to somebody else.

Finding reviews waiting for your input

Please remember to regularly check https://rfcbot.rs/. Click on any occurrence of your nickname to go to a page like https://rfcbot.rs/fcp/SimonSapin that only shows the reviews that are waiting for your input.

Reviewing PRs

As a member of the Libs team you’ll find yourself assigned to PRs that need reviewing, and your input requested on issues in the Rust project.

When is an RFC needed?

New unstable features don't need an RFC before they can be merged. If the feature is small, and the design space is straightforward, stabilizing it usually only requires the feature to go through FCP. Sometimes however, you may ask for an RFC before stabilizing.

Is there any unsafe?

Unsafe code blocks in the standard library need a comment explaining why they're ok. There's a tidy lint that checks this. The unsafe code also needs to actually be ok.

The rules around what's sound and what's not can be subtle. See the Unsafe Code Guidelines WG for current thinking, and consider pinging @rust-lang/libs, @rust-lang/lang, and/or somebody from the WG if you're in any doubt. We love debating the soundness of unsafe code, and the more eyes on it the better!

Is that #[inline] right?

Inlining is a trade-off between potential execution speed, compile time and code size. There's some discussion about it in this PR to the hashbrown crate. From the thread:

#[inline] is very different than simply just an inline hint. As I mentioned before, there's no equivalent in C++ for what #[inline] does. In debug mode rustc basically ignores #[inline], pretending you didn't even write it. In release mode the compiler will, by default, codegen an #[inline] function into every single referencing codegen unit, and then it will also add inlinehint. This means that if you have 16 CGUs and they all reference an item, every single one is getting the entire item's implementation inlined into it.

You can add #[inline]:

• To public, small, non-generic functions.

You shouldn't need #[inline]:

• On methods that have any generics in scope.
• On methods on traits that don't have a default implementation.

#[inline] can always be introduced later, so if you're in doubt they can just be removed.

What about #[inline(always)]?

You should just about never need #[inline(always)]. It may be beneficial for private helper methods that are used in a limited number of places or for trivial operators. A micro benchmark should justify the attribute.

Is there any potential breakage?

Breaking changes should be avoided when possible. RFC 1105 lays the foundations for what constitutes a breaking change. Breakage may be deemed acceptable or not based on its actual impact, which can be approximated with a crater run.

There are strategies for mitigating breakage depending on the impact.

For changes where the value is high and the impact is high too:

• Using compiler lints to try phase out broken behavior.

If the impact isn't too high:

• Looping in maintainers of broken crates and submitting PRs to fix them.

Is behavior changed?

Breaking changes aren't just limited to compilation failures. Behavioral changes to stable functions generally can't be accepted. See the home_dir issue for an example.

Are there new impls for stable traits?

A lot of PRs to the standard library are adding new impls for already stable traits, which can break consumers in many weird and wonderful ways. The following sections gives some examples of breakage from new trait impls that may not be obvious just from the change made to the standard library.

Inference breaks when a second generic impl is introduced

Rust will use the fact that there's only a single impl for a generic trait during inference. This breaks once a second impl makes the type of that generic ambiguous. Say we have:

#![allow(unused)]
fn main() {
// in `std`
impl From<&str> for Arc<str> { .. }
}

#![allow(unused)]
fn main() {
// in an external `lib`
let b = Arc::from("a");
}

then we add:

impl From<&str> for Arc<str> { .. }
+ impl From<&str> for Arc<String> { .. }

then

#![allow(unused)]
fn main() {
let b = Arc::from("a");
}

will no longer compile, because we've previously been relying on inference to figure out the T in Box<T>.

This kind of breakage can be ok, but a crater run should estimate the scope.

Deref coercion breaks when a new impl is introduced

Rust will use deref coercion to find a valid trait impl if the arguments don't type check directly. This only seems to occur if there's a single impl so introducing a new one may break consumers relying on deref coercion. Say we have:

#![allow(unused)]
fn main() {
// in `std`
impl Add<&str> for String { .. }

impl Deref for String { type Target = str; .. }
}

#![allow(unused)]
fn main() {
// in an external `lib`
let a = String::from("a");
let b = String::from("b");

let c = a + &b;
}

then we add:

impl Add<&str> for String { .. }
+ impl Add<char> for String { .. }

then

#![allow(unused)]
fn main() {
let c = a + &b;
}

will no longer compile, because we won't attempt to use deref to coerce the &String into &str.

This kind of breakage can be ok, but a crater run should estimate the scope.

Could an implementation use existing functionality?

Types like String are implemented in terms of Vec<u8> and can use methods on str through deref coersion. Vec<T> can use methods on [T] through deref coersion. When possible, methods on a wrapping type like String should defer to methods that already exist on their underlying storage or deref target.

Are there #[fundamental] items involved?

Blanket trait impls can't be added to #[fundamental] types because they have different coherence rules. See RFC 1023 for details. That includes:

• &T
• &mut T
• Box<T>
• Pin<T>

Is specialization involved?

Specialization is currently unstable. You can track its progress here.

We try to avoid leaning on specialization too heavily, limiting its use to optimizing specific implementations. These specialized optimizations use a private trait to find the correct implementation, rather than specializing the public method itself. Any use of specialization that changes how methods are dispatched for external callers should be carefully considered.

As an example of how to use specialization in the standard library, consider the case of creating an Rc<[T]> from a &[T]:

#![allow(unused)]
fn main() {
impl<T: Clone> From<&[T]> for Rc<[T]> {
    #[inline]
    fn from(v: &[T]) -> Rc<[T]> {
        unsafe { Self::from_iter_exact(v.iter().cloned(), v.len()) }
    }
}
}

It would be nice to have an optimized implementation for the case where T: Copy:

#![allow(unused)]
fn main() {
impl<T: Copy> From<&[T]> for Rc<[T]> {
    #[inline]
    fn from(v: &[T]) -> Rc<[T]> {
        unsafe { Self::copy_from_slice(v) }
    }
}
}

Unfortunately we couldn't have both of these impls normally, because they'd overlap. This is where private specialization can be used to choose the right implementation internally. In this case, we use a trait called RcFromSlice that switches the implementation:

#![allow(unused)]
fn main() {
impl<T: Clone> From<&[T]> for Rc<[T]> {
    #[inline]
    fn from(v: &[T]) -> Rc<[T]> {
        <Self as RcFromSlice<T>>::from_slice(v)
    }
}

/// Specialization trait used for `From<&[T]>`.
trait RcFromSlice<T> {
    fn from_slice(slice: &[T]) -> Self;
}

impl<T: Clone> RcFromSlice<T> for Rc<[T]> {
    #[inline]
    default fn from_slice(v: &[T]) -> Self {
        unsafe { Self::from_iter_exact(v.iter().cloned(), v.len()) }
    }
}

impl<T: Copy> RcFromSlice<T> for Rc<[T]> {
    #[inline]
    fn from_slice(v: &[T]) -> Self {
        unsafe { Self::copy_from_slice(v) }
    }
}
}

Only specialization using the min_specialization feature should be used. The full specialization feature is known to be unsound.

Are const generics involved?

Const generics are currently unstable. You can track their progress here.

Using const generics in public APIs is ok, but only const generics using the min_const_generics feature should be used publicly for now.

Are there public enums?

Public enums should have a #[non_exhaustive] attribute if there's any possibility of new variants being introduced, so that they can be added without causing breakage.

Does this change drop order?

Changes to collection internals may affect the order their items are dropped in. This has been accepted in the past, but should be noted.

Is there a manual Drop implementation?

A generic Type<T> that manually implements Drop should consider whether a #[may_dangle] attribute is appropriate on T. The Nomicon has some details on what #[may_dangle] is all about.

If a generic Type<T> has a manual drop implementation that may also involve dropping T then dropck needs to know about it. If Type<T>'s ownership of T is expressed through types that don't drop T themselves such as ManuallyDrop<T>, *mut T, or MaybeUninit<T> then Type<T> also needs a PhantomData<T> field to tell dropck that T may be dropped. Types in the standard library that use the internal Unique<T> pointer type don't need a PhantomData<T> marker field. That's taken care of for them by Unique<T>.

As a real-world example of where this can go wrong, consider an OptionCell<T> that looks something like this:

#![allow(unused)]
fn main() {
struct OptionCell<T> {
    is_init: bool,
    value: MaybeUninit<T>,
}

impl<T> Drop for OptionCell<T> {
    fn drop(&mut self) {
        if self.is_init {
            // Safety: `value` is guaranteed to be fully initialized when `is_init` is true.
            // Safety: The cell is being dropped, so it can't be accessed again.
            unsafe { self.value.assume_init_drop() };
        }
    }
}
}

Adding a #[may_dangle] attribute to this OptionCell<T> that didn't have a PhantomData<T> marker field opened up a soundness hole for T's that didn't strictly outlive the OptionCell<T>, and so could be accessed after being dropped in their own Drop implementations. The correct application of #[may_dangle] also required a PhantomData<T> field:

struct OptionCell<T> {
    is_init: bool,
    value: MaybeUninit<T>,
+   _marker: PhantomData<T>,
}

- impl<T> Drop for OptionCell<T> {
+ unsafe impl<#[may_dangle] T> Drop for OptionCell<T> {

How could mem break assumptions?

mem::replace and mem::swap

Any value behind a &mut reference can be replaced with a new one using mem::replace or mem::swap, so code shouldn't assume any reachable mutable references can't have their internals changed by replacing.

mem::forget

Rust doesn't guarantee destructors will run when a value is leaked (which can be done with mem::forget), so code should avoid relying on them for maintaining safety. Remember, everyone poops.

It's ok not to run a destructor when a value is leaked because its storage isn't deallocated or repurposed. If the storage is initialized and is being deallocated or repurposed then destructors need to be run first, because memory may be pinned. Having said that, there can still be exceptions for skipping destructors when deallocating if you can guarantee there's never pinning involved.

How is performance impacted?

Changes to hot code might impact performance in consumers, for better or for worse. Appropriate benchmarks should give an idea of how performance characteristics change. For changes that affect rustc itself, you can also do a rust-timer run.

Is the commit log tidy?

PRs shouldn’t have merge commits in them. If they become out of date with master then they need to be rebased.

Merging PRs

PRs to rust-lang/rust aren’t merged manually using GitHub’s UI or by pushing remote branches. Everything goes through bors.

When to rollup

For Libs PRs, rolling up is usually fine, in particular if it's only a new unstable addition or if it only touches docs.

See the rollup guidelines for more details on when to rollup. The idea is to try collect a number of PRs together and merge them all at once, rather than individually. This can get things merged faster, but might not be appropriate for some PRs that are likely to conflict, or have performance characteristics that would be obscured in a rollup.

When there's new public items

If the feature is new, then a tracking issue should be opened for it. Have a look at some previous tracking issues to get an idea of what needs to go in there. The issue field on #[unstable] attributes should be updated with the tracking issue number.

Unstable features can be merged as normal through bors once they look ready.

When there's new trait impls

There’s no way to make a trait impl for a stable trait unstable, so any PRs that add new impls for already stable traits must go through a FCP before merging. If the trait itself is unstable though, then the impl needs to be unstable too.

When a feature is being stabilized

Features can be stabilized in a PR that replaces #[unstable] attributes with #[stable] ones. The feature needs to have an accepted RFC before stabilizing. They also need to go through a FCP before merging.

You can find the right version to use in the #[stable] attribute by checking the Forge.

When a const function is being stabilized

Const functions can be stabilized in a PR that replaces #[rustc_const_unstable] attributes with #[rustc_const_stable] ones. The Constant Evaluation WG should be pinged for input on whether or not the const-ness is something we want to commit to. If it is an intrinsic being exposed that is const-stabilized then @rust-lang/lang should also be included in the FCP.

Check whether the function internally depends on other unstable const functions through #[allow_internal_unstable] attributes and consider how the function could be implemented if its internally unstable calls were removed. See the Stability attributes page for more details on #[allow_internal_unstable].

Where unsafe and const is involved, e.g., for operations which are "unconst", that the const safety argument for the usage also be documented. That is, a const fn has additional determinism (e.g. run-time/compile-time results must correspond and the function's output only depends on its inputs...) restrictions that must be preserved, and those should be argued when unsafe is used.

When a feature is being deprecated

To try reduce noise in the docs from deprecated items, they should be moved to the bottom of the module or impl block so they're rendered at the bottom of the docs page. The docs should then be cut down to focus on why the item is deprecated rather than how you might use it.

Release

This section documents the process around creating a new release of the compiler, tools, as well information on The Rust Programming Language's platform support.

External Links

• The Homu/Bors page provides links to the pull request testing queues for the rust-lang GitHub organisation, as well as providing an overview of the bot's syntax you can use to interact with it.
• Rustup Component History documents when a component was last available (if it was available) for a specific platform on nightly.
• PR Tracking provides visualisations of pull requests made to the rust-lang/rust repository.
• kennytm's rustup-toolchain-install-master is a utility to install the latest generated artifacts from CI into rustup.

Backporting

There's a steady trickle of patches that need to be ported to the beta and stable branch. Only a few people are even aware of the process, but this is actually something anybody can do.

Beta backporting in rust-lang/rust

Backports of PRs to the beta branch are usually only done to fix regressions. Getting a PR backported to the beta branch involves the following process:

1. Add the label beta-nominated to the PR to be backported. This marks the PR as in the state that it needs attention from the appropriate team to decide if it should be backported. Anybody with triage access is free to add this label.

2. If the team thinks it should be backported, then they should add the beta-accepted label. Otherwise they should remove the nominated label.

3. Occasionally someone will make a beta rollup PR. This is often done by the release team, but it can be done by anyone. The process here is:

   1. Create a local branch off the beta branch.

   2. Cherry-pick all of the PRs that have both beta-nominated and beta-accepted labels. It is usually preferred to not include PRs that have not been merged in case there are any last minute changes, or it fails when running the full CI tests.

   3. Run ./x.py run replace-version-placeholder and if there were any changes, put them into a new commit.

   4. (Recommended) Run some tests locally. It is not uncommon that the backports may not apply cleanly, or the UI tests need to be re-blessed if there are differences in the output.

   5. Open a PR against the beta branch with a title that starts with [beta] (so reviewers can see its specialness).

   6. List all of the PRs being backported in the PR description. Here's an example.

   7. Go through all of the PRs being backported and:

      • Change the milestone to the correct value for the beta release.
      • Remove the beta-nominated label. This indicates that the backport has been completed.

      If there are a lot of PRs, this can be done quickly by opening the nominated + accepted query, check all the PRs being backported, and use the "Milestones" and "Label" drop-downs to modify multiple PRs in bulk.

      This last step can be done before or after the beta PR has been merged, though it can be easy to forget if you wait for it to be merged.

4. A reviewer (typically from the release team) needs to verify that the backport looks correct and that it's submitted to the beta branch. They will then approve with @bors r+ rollup=never (to avoid it being rolled up on accident). If the author of the PR has r+ rights, and has not made significant changes while backporting, they can also self-approve the PR.

In summary, there are three states that a PR can go through:

1. beta-nominated: Needs the team's attention.
2. beta-nominated + beta-accepted: Waiting to be backported.
3. beta-accepted: Backport complete.

Stable backporting in rust-lang/rust

Backports to the stable branch work exactly the same as beta ones, labels have just a slightly different name: stable-nominated identifies a PR to be discussed for a backport and stable-accepted is a PR accepted for backport. Declined stable nomination will have the stable-nominated label removed.

The T-release will decide on a case by case basis if a stable backport will warrant a point (.patch) release (f.e. release a 1.50.1 between 1.50 and 1.51).

Beta Backporting in rust-lang/cargo

The procedure for backporting fixes to Cargo is similar but slightly more extended than the rust-lang/rust repo's procedure. Currently there aren't backport tags in the Cargo repository, but you'll initiate the backport process by commenting on an associated PR, requesting a backport. Once a Cargo team member has approved the backport to happen you're good to start sending PRs!

• First you'll send a PR to the rust-1.21.0 branch of Cargo (replace 1.21 with the current rustc beta version number). Like with rust-lang/rust you'll prefix the title of your PR with [beta] and ensure it's flagged as going to beta.

• Next a Cargo reviewer will @bors: r+ the PR and put it into the queue. Eventually bors will automatically merge the PR (when tests are passing) to the appropriate Cargo branch.

• Finally you'll send a PR to the rust-lang/rust repository's beta branch, updating the Cargo submodule. The Cargo submodule should be updated to the tip of the rust-1.21.0 branch (the branch your Cargo PR was merged to). As like before, ensure you've got [beta] in the PR title.

After that's all said and done the Cargo change is ready to get scheduled onto the beta release!

Preparing Release Notes

The release notes for the next release should be compiled at the beginning of the beta cycle, 6 weeks ahead of the release.

Clone the relnotes utility. This program pulls all pull requests made against rust-lang/rust and rust-lang/cargo within the latest release cycle and prints out a markdown document containing all the pull requests, categorised into their respective sections where possible, and prints the document to stdout.

Only pull requests that impact stable users of Rust should be included. Generally, more exciting items go toward the top of sections. Most items are simply links to the PR that landed them; some that need more explanation have additional, unlinked text; anything supported by an RFC has an additional RFC link. Reuse the PR titles or write descriptions as needed for clarity.

Try to keep the language of the document independent of en-US or en-UK, when it can't be avoided defer to en-US grammar and syntax.

The Rust Release Process

Here's how Rust is currently released:

Bump the stable version number (T-6 days, Friday the week before)

Open a PR bumping the version number in src/version. r+ rollup=never this PR.

Mark it as rollup=never, because if it lands in a rollup as not the first PR then other pull requests in that rollup will be incorrectly associated with the prior release.

This is effectively when the beta branch forks -- when beta is promoted, it will be based off of the PR that landed just before this version number bump PR.

Promote branches (T-3 days, Monday)

Both promotions should happen on Monday. You can open both PRs at the same time, but make sure the stable promotion lands first.

Beta to stable

Obtain AWS CLI credentials and run this command from the simpleinfra repository:

./release-scripts/promote-release.py branches

Once that's done, send a PR to the freshly created beta branch of rust-lang/rust with two commits:

• The changes caused by running ./x.py run replace-version-placeholder
• An update of src/ci/channel to beta

The version placeholder replacement changes must be in a separate commit so that they can be cherry picked to the master branch.

Also send a PR to rust-lang/rust targeting the new stable branch making the following changes:

• Update src/ci/channel to stable
• Update release notes to the latest available copy
  • e.g., git checkout origin/master -- RELEASES.md

Once the PRs are sent, r+ both and give them a high p=1000 (for stable) and p=10 for beta.

After the PR is merged you'll need to start a dev release. Obtain AWS CLI credentials and run this command from the simpleinfra repository:

# The date here is of the actual, production, stable release. Used for the blog post.
./release-scripts/promote-release.py release dev stable --release-date YYYY-MM-DD

Master bootstrap update (T-2 day, Tuesday)

Send a PR to the master branch to:

• Cherry pick the commit that ran ./x.py run replace-version-placeholder from the now merged beta branch PR. Do not re-run the tool as there might have been other stabilizations on master which were not included in the branched beta, so may not be attributed to the current release.

• Run ./x.py run src/tools/bump-stage0 to update the bootstrap compiler to the beta you created yesterday.

• Remove references to the bootstrap and not(bootstrap) conditional compilation attributes. You can find all of them by installing ripgrep and running this command:

  rg '#!?\[.*\(bootstrap' -t rust
  

  The general guidelines (both for #[] and #![]) are:

  • Remove any item annotated with #[cfg(bootstrap)].
  • Remove any #[cfg(not(bootstrap))] attribute while keeping the item.
  • Remove any #[cfg_attr(bootstrap, $attr)] attribute while keeping the item.
  • Replace any #[cfg_attr(not(bootstrap), doc="$doc")] with $doc in the relevant documentation block (or in a new documentation block).
  • Replace any #[cfg_attr(not(bootstrap), $attr)] with #[$attr].

Release day (Thursday)

Decide on a time to do the release, T.

• T-50m - Run the following command in a shell with AWS credentials in the simpleinfra repository:

  ./release-scripts/promote-release.py release prod stable
  

  That'll, in the background, schedule the promote-release binary to run on the production secrets (not the dev secrets). That'll sign everything, upload it, update the html index pages, and invalidate the CDN. Note that this takes about 30 minutes right now. This will also push a signed tag to rust-lang/rust.

• T-2m - Merge blog post.

• T - Tweet and post everything!

  • Twitter @rustlang
  • Users forum

• T+5m - Release and tag Cargo. From a rust-lang/rust checkout (script will checkout the stable branch automatically), run the following script from simpleinfra.

  ../simpleinfra/release-scripts/tag-cargo.sh
  

• T+1hr Send a PR to the beta branch running ./x.py run src/tools/bump-stage0 to bump the boostrap compiler to the stable you just released.

Bask in your success.

Rebuilding stable pre-releases

If something goes wrong and we need to rebuild the stable artifacts, merge the PR on the stable branch of the rust-lang/rust repository. Once the commit is merged, issue the following command in a shell with AWS credentials on the simpleinfra repository:

./release-scripts/promote-release.py release dev stable --bypass-startup-checks

You'll also want to update the previously published blog post and internals post with the new information.

Publishing a nightly based off a try build

Sometimes a PR requires testing how it behaves when downloaded from rustup, for example after a manifest change. In those cases it's possible to publish a new nightly based off that PR on dev-static.rust-lang.org.

Once the try build finishes grab the merge commit SHA and run the following command in a shell with AWS credentials on the simpleinfra repository:

./release-scripts/promote-release.py release dev nightly $MERGE_COMMIT_SHA

When the release process end you'll be able to install the new nightly with:

RUSTUP_DIST_SERVER=https://dev-static.rust-lang.org rustup toolchain install nightly

Rollup Procedure

Background

The Rust project has a policy that every pull request must be tested after merge before it can be pushed to master. As PR volume increases this can scale poorly, especially given the long (~3.5hr) current CI duration for Rust.

Enter rollups! Changes that are small, not performance sensitive, or not platform dependent are marked with the rollup command to bors (@bors r+ rollup to approve a PR and mark as a rollup, @bors rollup to mark a previously approved PR, @bors rollup- to un-mark as a rollup). 'Performing a Rollup' then means collecting these changes into one PR and merging them all at once. The rollup command accepts four values always, maybe, iffy, and never. See the Rollups section of the review policies for guidance on what these different statuses mean.

You can see the list of rollup PRs on Rust's Homu queue, they are listed at the bottom of the 'approved' queue with a priority of 'rollup' meaning they will not be merged by themselves until everything in front of them in the queue has been merged.

Making a Rollup

1. Using the interface on Homu queue, select pull requests and then use "rollup" button to make a rollup pull request. (The text about fairness can be ignored.) Important note: consider for addition PRs marked as rollup=always, rollup=maybe and rollup=iffy, based on the review policies of the Rollups section. Be extra careful when deciding what to include, in particular on rollup=maybe and rollup=iffy PRs. We should try as much as possible to avoid risking and hit regressions (bugs or perf). Also consider that contributors often forget to tag things with rollup=never, when they should have done so, so when PRs are not explicitly tagged with rollup, be extra careful.

2. Run the following command in the pull request thread:

   @bors r+ rollup=never p=5
   

3. If the rollup fails, use the logs rust-log-analyzer provides to bisect the failure to a specific PR and do @bors r-. If the PR is running, you need to do @bors r- retry. Otherwise, your rollup succeeded. If it did, proceed to the next rollup (every now and then let rollup=never and toolstate PRs progress).

4. Recreate the rollup without the offending PR starting again from 1.. There's a link in the rollup PR's body to automatically prefill the rollup UI with the existing PRs (minus any PRs that have been r-d)

Selecting Pull Requests

The queue is sorted by rollup status. In general, a good rollup includes one or two iffy PRs (if available), a bunch of maybe (unmarked) PRs, and a large pile of always PRs. A rollup should never include rollup=never PRs.

The actual absolute size of the rollup can depend based on experience, people new to making rollups might start with including 1 iffy, 4 maybes, and 5 alwayss, but more experienced people might even make a rollup of 1-2 iffys, 8 maybes, and 10 alwayss! Massive rollups are rarely needed, but as your intuition grows you'll get better at judging risk when including PRs in a rollup.

Don't hesitate to downgrade the rollup status of a PR! If your intuition tells you that a rollup=always PR has some chances for failures, mark it rollup=maybe or rollup=iffy. A lot of the unmarked maybe PRs are categorized as such because the reviewer may not have considered rollupability, so it's always worth picking them with a critical eye. Similarly, if a PR causes your rollup to fail, it's worth considering changing its rollup status

Generally, PRs, that touch CI configuration or the bootstrapping process are probably iffy and should be handled with care. On the other hand, PRs that just edit docs are usually rollup=always.

Avoid having too many PRs with large diffs or submodule changes in the same rollup. Also avoid having PRs you suspect will have large perf impacts, and mark them as rollup=never.

It's tempting to avoid including iffy PRs at all since ideally you want your rollup to succeed. However, it's worth remembering that the job of the PR queue is to test PRs, not to land them. As such, a rollup that fails because of an iffy PR is a good thing, since that PR would have to be tested at some point anyway and it would have taken up the same amount of time to test if it never got included in a rollup. One way to look at rollups when it comes to iffy PRs is that a rollup is a way for a bunch of other PRs to piggyback on the CI cycle that the iffy PR needs anyway. If rollups avoid iffy PRs entirely what ends up happening is that these PRs tend to languish in the queue for a long time, which isn't good.

Similarly, make sure to leave some spare CI cycles so that never PRs also get a chance! If you're the only person making rollups it's worth letting them run during times you're not paying attention to the queue, but these days there are rollup authors in multiple time zones, so it's often best to just keep an eye on the relative size of the queue and put aside a couple CI cycles for never PRs, especially if they pile up.

Try to be fair with rollups: Rollups are a way for things to jump the queue. For rollup=maybe PRs, try to include the oldest one (at the top of the section) so that newer PRs aren't jumping the queue over older PRs entirely. You don't have to include every PR older than PRs included in your rollup, but try to include the oldest. Similar to the perspective around iffy, it's useful to look at a rollup as a way for other PRs to piggyback on the CI cycle of the oldest PR in queue.

Failed rollups

If the rollup has failed, run the @bors retry command if the failure was spurious (e.g. due to a network problem or a timeout). If it wasn't spurious, find the offending PR and throw it out by copying a link to the rust-logs-analyzer comment, and writing Failed in <link_to_comment>, @bors r-. Hopefully, the author or reviewer will give feedback to get the PR fixed or confirm that it's not at fault. The failed rollup PR can be closed.

Once you've removed the offending PR, re-create your rollup without it (see 1.). Sometimes however, it is hard to find the offending PR. If so, use your intuition to avoid the PRs that you suspect are the problem and recreate the rollup. Another strategy is to raise the priority of the PRs you suspect, mark them as rollup=never (or iffy) and let bors test them standalone to dismiss or confirm your hypothesis.

If a rollup continues to fail you can run the @bors rollup=never command to never rollup the PR in question.

Triage Procedure

Pull Request Triage

Status Tags

• S-waiting-on-author - Author needs to make changes to address reviewer comments, or merge conflicts/test failures are present. This also covers more obscure cases, like a PR being blocked on another (usually with the S-blocked label in addition), or waiting for a crater run -- it is the author's responsibility to push the PR forward.

  Also used for work-in-progress PRs, sometimes the PR will also be marked as draft in GitHub.

• S-waiting-on-review - Review is incomplete

• S-waiting-on-team - A T- label is marked, and team has been CC'd for feedback.

• S-waiting-on-bors - Currently approved, waiting to merge. Managed by bors.

• S-waiting-on-crater - Waiting to see what the impact the PR will have on the ecosystem

• S-waiting-on-bikeshed - Waiting on the consensus over a minor detail

• S-waiting-on-perf - Waiting on the results of a perf run

• S-blocked - Waiting for another PR to be merged or for discussion to be resolved

• S-inactive - Hasn't had activity in a while

• S-experimental - An experimental PR that shouldn't be triaged. S-waiting-on-author used to be used for this, but S-experimental communicates that the PR is an experiment to test out some changes.

Also: PRs with no status tags. This is useful to find PRs where rustbot conked out and didn't assign a reviewer and thus didn't assign S-waiting-on-review. These PRs can get lost otherwise. (Note that you should likely not triage PRs that have r? @ghost since that means the author does not need a review yet.)

Procedure

We primarily triage three status labels: S-waiting-on-review, S-waiting-on-author, and (once in a while) S-blocked. Here is the procedure for each:

S-waiting-on-review

Click this link to see all PRs with the S-waiting-on-review label. Only triage PRs that were last updated 15 days or more ago (give or take a day).

For each PR:

1. If the PR has new conflicts, CI failed, or a new review has been made then change the label to S-waiting-on-author and ping the author.

2. Add the PR to your report.

S-waiting-on-author

Click this link to see all PRs with the S-waiting-on-author label. Only triage PRs that were last updated 15 days or more ago (give or take a day).

For each PR:

1. If the author did what the PR was waiting on them for then update the label to S-waiting-on-review.

   Otherwise, if the author still needs to do something, then ping the author if they are not a member of a Rust team (does not include working groups — only teams like T-compiler, T-lang, T-rustdoc, etc.).

2. Add the PR to your report.

S-blocked

You only need to check S-blocked PRs occasionally (e.g., once a month). Click this link to see all PRs with the S-blocked label.

For each PR:

1. If it is still blocked then leave it as-is.

   Otherwise, if it is no longer blocked, then remove S-blocked (and add a status label like S-waiting-on-review if appropriate).

2. Add the PR to your report.

Triage Report

You should record information about each PR you triage in a report. The report is just a small document that looks like:

S-waiting-on-review

#12345 20 days - still waiting on review - author: ferris, assignee: bors

[...]

Your report can look different, just make sure you include this information for each PR:

1. The PR number (e.g., #12345). No need to manually add a link; the Rust Zulip will autolink PR (and issue) numbers.

2. Number of days since last activity. "Activity" means:

   • author, reviewer, or team member commented or reviewed; or
   • bors commented about merge conflicts; or
   • PR was pushed to;
   • etc.

3. Author, reviewer, and who or what (person, team, other PR, etc.) the PR is waiting on.

4. Current status and what the most recent activity was (e.g., merge conflicts, reviewer commented).

Once you are done triaging PRs, post your report in the topic for the current week's triage in the #t-release/triage Zulip stream.

Triaging Crater Runs

Running crater

We regularly run Crater runs, and this documents the procedure for triaging a beta run; it may also be applicable to non-release team runs (e.g., PR crater runs) with minor modifications.

First, file a new issue titled "Crater runs for 1.x" (example)

A crater run for beta should be started as soon as we have beta out. Use the following craterbot invocations.

$BETA_VERSION is e.g. 1.40.0-1, increment the 1 if it's not the first beta crater run, you can also use the auto-incremented counter on the beta rustc --version.

$STABLE is e.g. 1.39.0 (the stable release) $BETA is beta-YYYY-MM-DD, get the date by looking at https://static.rust-lang.org/manifests.txt and get the date of the most recent channel-rust-beta.toml.

@craterbot run name=beta-$BETA_VERSION start=$STABLE end=$BETA mode=build-and-test cap-lints=warn p=10
@craterbot run name=beta-rustdoc-$BETA_VERSION start=$STABLE end=$BETA mode=rustdoc cap-lints=warn p=5

Once the runs complete, you want to triage them

Triaging

These steps should generally be done for the normal rustc run, and then followed up by a triage of the rustdoc run. Ignore failures in rustdoc that look to be rooted in rustc (i.e., duplicate failures).

There will usually be quite a few regressions -- there are a couple tools that can help reduce the amount of work that you need to do. It's mostly a matter of personal preference which is more helpful.

• https://github.com/Mark-Simulacrum/crater-generate-report/
  • This groups regressions by 'root' by parsing the logs to look for the compilation failed messages printed by Cargo
• https://github.com/Centril/crater-cat-errors
  • This groups regressions by the "error" message, also by parsing logs

If you've written a tool, feel free to add it here! We're still figuring out what the best UI for this is.

Regardless of the tool you've run, you ultimately need to read through a bunch of logs and try to quickly determine if they're genuine failures or spurious. Most of the time, a compiler failure is genuine, and test failures are mostly spurious, but this usually requires some level of guessing.

Once you've determined that something is a genuine failure, add it to a list somewhere (local file, HackMD, whatever) with the error "category." Mostly, you're trying to group things such that the regressions in a single group are all caused by the same set of commits, and different groups have different causes.

Once this is done, and you have all the regressions triaged into their separate groups, you want to file a new issue for each group. It should have the regression-from-stable-to-beta and T-compiler label by default, possibly T-libs if it's a standard library regression, but that's relatively rare. If you happen to think you know the PR that caused the failure, cc the PR author in a separate comment and link to the PR; otherwise compiler team will triage the issue soon.

Leave a comment on the original issue with the crater runs linking to all the issues you just opened, ideally with the issue titles as well.

You're done!

Re-running rustc on a crate

For the crates which we're not sure about, you can try running crater locally, or build the crate directly (cratesio-curl can be helpful). Be careful -- regardless of what you do, you are running arbitrary code locally. It's also fine to file issues for the crates you're not sure about and let the triage process naturally categorize the error, though it's not good to do this for all the crates. Once you've triaged a crater run a couple times you get a pretty good sense of what is spurious and what isn't, too.

You can run crater on just a single crate by doing something like this (at least, as of now). Note that this will download several gigabytes (on first use) and requires Docker to be running.

git clone https://github.com/rust-lang/crater
cd crater
cargo run -- prepare-local
CRATES="crates-io-crate-0.4.0,owner/repository-name" # Edit this.
cargo run -- define-ex --crate-select=list:$CRATES --cap-lints=forbid 1.38.0 beta # Edit the stable version.
cargo run -- run-graph --threads 4
cargo run -- gen-report work/ex/default/
# view report for this crate

It's also possible to re-queue a subset of crates onto the official builders, for which that take a look at: https://gist.github.com/ecstatic-morse/be799bfa4d3b3d6e163fa61a9c30706f

Determining the root cause of the regression

It's not always apparent why a crate stopped building. This isn't generally something done as part of crater triage -- but can be a good followup. Here, cargo-bisect-rustc and Felix's minimization guide are excellent tools to apply.

Archive

This section is for content that has become outdated, but that we want to keep available to be read for historical/archival reasons.

Friends of the Tree

The Rust Team likes to occasionally recognize people who have made outstanding contributions to The Rust Project, its ecosystem, and its community. These people are 'Friends of the Tree', archived here for eternal glory.

2016-02-26 @mitaa

This week we would like to nominate @mitaa as Friend of the Tree. Recently @mitaa has sent a wave of fixes to rustdoc (yes those are all separate links) with even more on the way! Rustdoc has historically been a tool in need of some love, and the extra help in fixing bugs is especially appreciated. Thanks @mitaa!

2016-02-12 Jeffrey Seyfried (@jseyfried)

This week's friend of the tree is Jeffrey Seyfried (@jseyfried)!

Jeffrey Seyfried (@jseyfried) has made some awesome contributions to name resolution. He has fixed a ton of bugs, reported previously unknown edge cases, and done some big refactorings, all of which have helped improve a complex and somewhat neglected part of the compiler.

2015-12-04 Vadim Petrochenkov @petrochenkov

This week we'd like to nominate @petrochenkov for Friend of the Tree. Vadim has been doing some absolutely amazing compiler work recently such as fixing privacy bugs, fixing hygiene bugs, fixing pattern bugs, paving the way and implementing #[deprecated], fixing and closing many privacy holes, refactoring and improving the HIR, and reviving the old type ascription PR. The list of outstanding bugs and projects in the compiler is growing ever smaller now; thanks @petrochenkov!

2015-11-16 Peter Atashian (WindowsBunny, retep998)

In his own words, WindowsBunny is "a hopping encyclopedia of all the issues windows users might run into and how to solve them." One of the heroes that make Rust work on Windows, he actively pushes the frontiers of what Rust can do on the platform. He is also notably the maintainer of the winapi family of crates, a comprehensive set of bindings to the Windows system APIs. You da bunny, WindowsBunny. Also, a friend of the tree.

Source.

2015-10-31 Marcus Klaas

Today @nrc would like to nominated @marcusklaas as Friend of the Tree:

Marcus is one of the primary authors of rustfmt. He has been involved since the early days and is now the top contributor. He has fixed innumerable bugs, implemented new features, reviewed a tonne of PRs, and contributed to the design of the project. Rustfmt would not be the software it is today without his hard work; he is indeed a Friend Of The Tree.

2015-10-16 Ryan Prichard

nmatsakis would also like to declare Ryan Prichard a Friend of the Tree. Over the last few months, Ryan has been comparing the Rust compiler's parsing behavior with that of the rust-grammar project, which aims to create a LALR(1) grammar for parsing Rust. Ryan has found a number of inconsistencies and bugs between the two. This kind of work is useful for two reasons: it finds bugs, obviously, which are often hard to uncover any other way. Second, it helps pave the way for a true Rust reference grammar outside of the compiler source itself. So Ryan Prichard, thanks!

2015-10-02 Vikrant Chaudhary

Vikrant Chaudhary (nasa42) is an individual who believes in the Rust community. Since June he has been contributing to This Week in Rust, coordinating its publication on urlo, and stirring up contributions. He recently rolled out an overhaul to the site's design that brings it more inline with the main website. Today Vikrant is the main editor on the weekly newsletter, assisted by llogiq and other contributors. Thanks for keeping TWiR running, Vikrant, you friend of the tree.

Source.

2015-07-24 Tshepang Lekhonkhobe

@Gankra has nominated @tshepang for Friend of the Tree this week:

Over the last year Tshepang has landed over 100 improvements to our documentation. Tshepang saw where documentation was not, and said "No. This will not do."

We should all endeavor to care about docs as much as Tshepang.

Source.

2015-05-19 Chris Morgan

I'd like to nominate Chris Morgan (@chris-morgan) for Friend of the Tree today. Chris recently redesigned the play.rust-lang.org site for the 1.0 release, giving the site a more modern and rustic feel to it. Chris has been contributing to Rust for quite some time now, his first contribution dating back to July 2013 and also being one of the early pioneers in the space of HTTP libraries written in Rust. Chris truly is a friend of the tree!

2015-03-24 Andrew Gallant (BurntSushi)

BurntSushi is an individual who practically needs no introduction. He's written many of the world's most popular crates, including docopt.rs, regex, quickcheck, cbor, and byteorder. Don't forget his CSV swiss-army-knife, xsv, built on rust-csv. Feedback from his early work on libraries helped informed the evolution of Rust during a critical time in its development, and BurntSushi continues to churn out the kind of Rust gems that can only come from someone who is a skilled friendofthetree.

2015-03-03 Manish Goregaokar (Manishearth)

Manish started working on Servo as part of the GSoC program in 2014, where he implemented XMLHttpRequest. Since then he's become in integral part of the Servo team while finishing his university studies and organizing Rust community events. In 2015 he took an interest in bors' queue and started making rollup PRs to accelerate the integration process. Nursing the PR queue is the kind of time-consuming labor that creates friends of the tree like Manish, the rollup friend of the tree.

2015-02-17 Toby Scrace

Today I would like to nominate Toby Scrace as Friend of the Tree. Toby emailed me over the weekend about a login vulnerability on crates.io where you could log in to whomever the previously logged in user was regardless of whether the GitHub authentication was successful or not. I very much appreciate Toby emailing me privately ahead of time, and I definitely feel that Toby has earned becoming Friend of the Tree.

2015-02-10 Jonathan Reem (reem)

Jonathan Reem has been making an impact on Rust since May 2014. His primary contribution has been as the main author of the prominent Iron web framework, though he has also created several other popular projects including the testing framework stainless. His practical experience with these projects has led to several improvements in upstream rust, most notably his complete rewrite of the TaskPool type. Reem is doing everything he can to advance the Rust cause.

2015-01-20 Barosl Lee (barosl)

Today I would like to nominate Barosl Lee (@barosl) for Friend of the Tree. Barosl has recently rewritten our bors cron job in a new project called homu. Homu has a number of benefits including:

• Zero "down time" between testing different PRs (compared to 30+ minutes for bors!)
• A new rollup button to create separate rollup PRs from other PRs.
• Multiple repositories are supported (Cargo and Rust are on the same page)

Homu was recently deployed for rust-lang/rust thanks to a number of issues being closed out by Barosl, and it's been working fantastically so far! Barosl has also been super responsive to any new issues cropping up. Barosl truly is a Friend of the Tree!

2015-01-13 Kang Seonghoon (lifthrasiir, Yurume)

Seonghoon has been an active member of the Rust community since early 2013, and although he has made a number of valuable contributions to Rust itself, his greatest work has been in developing key libraries out of tree. rust-encoding, one of the most popular crates in Cargo, performs character encoding, and rust-chrono date / time handling, both of which fill critical holes in the functionality of the standard library. rust-strconv is a prototype of efficient numerical string conversions that is a candidate for future inclusion in the standard library. He maintains a blog where he discusses his work.

2015-01-06 Jorge Aparicio (japaric)

I nominate Jorge Aparicio (japaric) for Friend of the Tree (for the second time, no less!). japaric has done tremendous work porting the codebase to use the new language features that are now available. First, he converted APIs in the standard library to take full advantage of DST after it landed. Next, he converted APIs to use unboxed closures. Then, he converted a large portion of the libraries to use associated types. Finally, he removed boxed closures from the compiler entirely. He has also worked to roll out RFCs changing the overloaded operators and comparison traits, including both their definitions and their impact on the standard library. And this list excludes a number of smaller changes, like deprecating older syntax. The alpha release would not be where it is without him; Japaric is simply one of the best friends the tree has ever had.

2014-12-30 Kevin Ballard (kballard, Eridius)

This is a belated recognition of Kevin Ballard (aka @kballard, aka Eridius) as a friend of the tree. Kevin put a lot of work into Unicode issues in Rust, especially as related to platform-specific constraints. He wrote the current path module in part to accommodate these constraints, and participated in the recent redesign of the module. He has also been a dedicated and watchful reviewer. Thanks, Kevin, for your contributions!

2014-12-16 Gábor Lehel (glaebhoerl)

Gabor's major contributions to Rust have been in the area of language design. In the last year he has produced a number of very high quality RFCs, and though many of them of not yet been accepted, his ideas are often thought-provoking and have had a strong influence on the direction of the language. His trait based exception handling RFC was particularly innovative, as well that for future-proofing checked arithmetic. Gabor is an exceedingly clever Friend of the Tree.

2014-11-11 Brian Koropoff (unwound)

In the last few weeks, he has fixed many, many tricky ICEs all over the compiler, but particularly in the area of unboxed closures and the borrow checker. He has also completely rewritten how unboxed closures interact with monomorphization and had a huge impact on making them usable. Brian Koropoff is truly a Friend of the Tree.

2014-10-07 Alexis Beingessner (Gankra)

Alexis Beingessner (aka @Gankra) began contributing to Rust in July, and has already had a major impact on several library-related areas. Her main focus has been collections. She completely rewrote BTree, providing a vastly more complete and efficient implementation. She proposed and implemented the new Entry API. She's written extensive new documentation for the collections crate. She pitched in on collections reform.

And she added collapse-all to rustdoc!

Alexis is, without a doubt, a FOTT.

2014-09-02 Jorge Aparicio (japaric)

Jorge has made several high-impact contributions to the wider Rust community. He is the primary author of rustbyexample.com, and last week published "eulermark", a comparison of language performance on project Euler problems, which happily showed Rust performing quite well. As part of his benchmarking work he has ported the 'criterion' benchmarking framework to Rust.

2014-07-29 Björn Steinbrink (dotdash, doener)

Contributing since April 2013. Björn has done many optimizations for Rust, including removing allocation bloat in iterators, fmt, and managed boxes; optimizing fail!; adding strategic inlining in the libraries; speeding up data structures in the compiler; eliminating quadratic blowup in translation, and other IR bloat problems.

He's really done an amazing number of optimizations to Rust.

Most recently he earned huge kudos by teaching LLVM about the lifetime of variables, allowing Rust to make much more efficient use of the stack.

Björn is a total FOTT.

2014-07-22 Jonas Hietala (treeman)

Jonas Hietala, aka @treeman, has been contributing a large amount of documentation examples recently for modules such as hashmap, treemap, priority_queue, collections, bigint, and vec. He has also additionally been fixing UI bugs in the compiler such as those related to format!

Jonas continues to add new examples/documentation every day, making documentation more approachable and understandable for all newcomers. Jonas truly is a friend of the tree!

2014-07-08 Sven Nilson (bvssvni, long_void)

Sven Nilson has done a great deal of work to build up the Rust crate ecosystem, starting with the well-regarded rust-empty project that provides boilerplate build infrastructure and - crucially - integrates well with other tools like Cargo.

His Piston project is one of the most promising Rust projects, and its one that integrates a number of crates, stressing Rust's tooling at just the right time: when we need to start learning how to support large-scale external projects.

Sven is a friend of the tree.

2014-06-24 Jakub Wieczorek (jakub-)

jakub-, otherwise known as Jakub Wieczorek, has recently been working very hard to improve and fix lots of match-related functionality, a place where very few dare to venture! Most of this code appears to be untouched for quite some time now, and it's receiving some well-deserved love now.

Jakub has fixed 10 bugs this month alone, many of which have been long-standing problems in the compiler. He has also been very responsive in fixing bugs as well as triaging issues that come up from fun match assertions.

Jakub truly is a friend of the tree!

2014-04-22 klutzy

klutzy has been doing an amazing amount of Windows work for years now. He picks up issues that affect our quality on Windows and picks them off 1 by 1. It's tedious and doesn't get a ton of thanks, but is hugely appreciated by us. As part of the Korean community, he has also done a lot of work for the local community there. He is a friend of the tree. Thank you!

• Rust on Windows crusader
• Fixed issues with x86 C ABI struct arguments
• Fixed multiple issues with non-US locales

2014-03-18 Clark Gaebel (cgaebel)

This week's friend of the tree is Clark Gaebel. He just landed a huge first contribution to Rust. He dove in and made our hashmaps significantly faster by implementing Robin Hood hashing. He is an excellent friend of the tree.

2014-02-25 Erick Tryzelaar (erickt)

• Contributing since May 2011
• Wrote the serialization crate
• Organizes the bay area Rust meetups
• Just rewrote the Hash trait

2014-02-11 Flavio Percoco (FlaPer87)

• Contributing since September
• Does issue triage
• Organizing community events in Italy
• Optimized the 'pow' function
• Recently been fixing lots of small but important bugs

2014-01-27 - Jeff Olson (olsonjefferey)

• Contributing since February 2012
• Did the original libuv integration
• Implemented our second attempt at I/O, first using libuv
• Ported parts of the C++ runtime to Rust
• Implemented file I/O for the newest runtime
• Last week published an article about file I/O on the Safari books blog

2014-01-21 - Steven Fackler (sfackler)

• Contributing since last May
• CMU grad
• Lots of library improvements, Base64, Bitv, I/O
• Rustdoc improvements
• Mut/RefCell
• std::io::util
• external module loading

2014-01-14 - Eduard Burtescu (eddyb)

• Contributing since October
• Working on the compiler, including trans
• Reduced rustc memory usage
• Optimized vector operations
• Helping refactor the compiler to eliminate use of deprecated features
• Cleaned up ancient code in the compiler
• Removed our long-standing incorrect use of the environment argument to pass the self param

2014-01-07 - Vadim Chugunov (vadimcn)

• Contributing since June
• Fixed numerous bugs on Windows
• Fixing broken tests
• Improved compatibility with newer mingw versions
• Eliminated our runtime C++ dependency by implementing unwinding through libunwind

Rust Release history

This is an archive of Rust release artifacts from 0.1–1.7.0. Each release is signed with the Rust GPG signing key (older key, even older key).

1.7.0

• Announcement
• Release notes
• Source code (signature)
• Windows x86_64 .exe gnu installer (signature)
• Windows x86_64 .msi gnu installer (signature)
• Windows x86_64 .exe MSVC installer (signature)
• Windows x86_64 .msi MSVC installer (signature)
• Windows i686 .exe gnu installer (signature)
• Windows i686 .msi gnu installer (signature)
• Windows i686 .exe MSVC installer (signature)
• Windows i686 .msi MSVC installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)

1.6.0

• Announcement
• Release notes

1.5.0

• Announcement
• Release notes

1.4.0

• Announcement
• Release notes

1.3.0

• Announcement
• Release notes

1.2.0

• Announcement
• Release notes

1.1.0

• Announcement
• Release notes

1.0.0

• Announcement
• Release notes

1.0.0-beta

• Announcement

1.0.0-alpha.2

• Announcement
• Release notes
• Source code (signature)
• Windows x86_64 .exe installer (signature)
• Windows i686 .exe installer (signature)
• Windows x86_64 .msi installer (signature)
• Windows i686 .msi installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X x86_64 tarball (signature)
• Mac OS X i686 tarball (signature)
• Documentation

1.0.0-alpha

• Announcement
• Release notes
• Source code (signature)
• Windows x86_64 installer (signature)
• Windows i686 installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X x86_64 tarball (signature)
• Mac OS X i686 tarball (signature)
• Documentation

Rust 0.x

In addition to the included short-form release in the mailing list, each 0.x release has a longer explanation in the release notes.

0.12.0

• Announcement
• Release notes
• Source code (signature)
• Windows x86_64 installer (signature)
• Windows i686 installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X x86_64 tarball (signature)
• Mac OS X i686 tarball (signature)
• Documentation

0.11.0

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X x86_64 tarball (signature)
• Mac OS X i686 tarball (signature)
• Documentation

0.10

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Linux x86_64 tarball (signature)
• Linux i686 tarball (signature)
• Mac OS X x86_64 pkg (signature)
• Mac OS X i686 pkg (signature)
• Mac OS X x86_64 tarball (signature)
• Mac OS X i686 tarball (signature)
• Documentation

0.9

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Documentation

0.8

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
  • borrowed pointers | conditions | containers | ffi | macros | rustpkg | tasks
• Manual (PDF)
• Rustpkg manual
• Standard library docs
• Extra library docs

0.7

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
• Manual (PDF)
• Standard library docs
• Extra library docs

0.6

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
• Manual (PDF)
• Core library docs
• Standard library docs

0.5

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
• Manual (PDF)
• Core library docs
• Standard library docs

0.4

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
• Manual (PDF)
• Core library docs
• Standard library docs

0.3.1

• Announcement
• Release notes
• Source code (signature)

This was an OS X bugfix release.

0.3

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)
• Tutorial
• Manual (PDF)
• Core library docs
• Standard library docs

0.2

• Announcement
• Release notes
• Source code (signature)
• Windows installer (signature)

0.1

• Announcement
• Release notes
• Source code (signature)