Contributing to the Rust compiler can be a daunting prospect for some. A large code-base, a complex build system and endless tickets means finding a nice starting place can be tough. This article aims to shed light on some of the areas you should be aware of when contributing your first patch to the Rust compiler, and provide some useful links to find more information.
There are some areas you need to be aware of to effectively work inside the Rust project. These are:
The build system
The build system for Rust is complex. It covers bootstrapping the compiler, running tests, building documentation and more. Unless you are familiar with Makefiles, I wouldn't suggest trying to understand everything going on in Rust's setup - there's a lot there, and you can get lost trying to understand it all.
If Makefiles are your thing, though, all the configuration lives in the
mk directory in the project root.
Before you can start building the compiler you need to configure the build for your system. In most cases, that will just mean using the defaults provided for Rust. Configuring involves invoking the
configure script in the project root.
There are large number of options accepted by this script to alter the configuration used later in the build process. Some options to note:
--enable-debug- Build a debug version of the compiler (disables optimizations)
--disable-valgrind-rpass- Don't run tests with valgrind
--enable-clang- Prefer clang to gcc for building dependencies (ie LLVM)
--enable-ccache- Invoke clang/gcc with ccache to re-use object files between builds
To see a full list of options, run
Building the Rust compiler from source involves four stages.
Because the source for the Rust compiler is itself written in Rust, it means that we can't just compile the source into the latest compiler. Instead, we need to download an older version of the compiler from the internet and use that to build a new version of the compiler from the source tree.
stage0 downloads an older version of the compiler from the internet.
Once we have a version of the compiler downloaded, we start to compile a new version from source. We use the compiler from stage0 to build the stage1 compiler.
The stage1 compiler contains all new language features & optimizations, but are not used in the compiler itself. To build the compiler with these new features & optimizations, we need to build from source again using this version of the compiler.
Using the compiler from stage1 which contains all new language features & optimizations we build the stage2 compiler. At this point, we have the latest, most advanced & optimized version of the compiler.
The final (and optional) build step is to re-build the compiler once more using the compiler from stage2 to produce the stage3 compiler.
The stage3 compiler should be bitwise identical to the stage2 compiler. We build it to ensure that we haven't introduced any new issues in the latest build.
The most common targets I find myself using are:
make rustc-stage1- build up to (and including) the first stage. For most cases we don't need to build the stage2 compiler, so we can save time by not building it. The stage1 compiler is a fully functioning compiler and (probably) will be enough to determine if your change works as expected.
make check- build the full compiler & run all tests (takes a while). This is what gets run by the continuous integration system against your pull request. You should run this before submitting to make sure your tests pass & everything builds in the correct manner.
make check-stage1-std NO_REBUILD=1- test the standard library without rebuilding the entire compiler
make check TESTNAME=<path-to-test-file>.rs- Run a single test file
make check-stage1-rpass TESTNAME=<path-to-test-file>.rs- Run a single rpass test with the stage1 compiler (this will be quicker than running the command above as we only build the stage1 compiler, not the entire thing). You can also leave off the
-rpassto run all stage1 test types.
Once you have your local environment set up you need to find a task to work on. Rust uses GitHub issues to track tasks.
The Rust team uses a series of issue labels to help sort through and categorize issues. These labels denote the area of the code-base to which this issue is related, the expected level of difficulty, priority, and so on.
All labels have a prefix which denote the label type.
A-means this label indicates the Area of the code-base related to this ticket. For example,
A-parsermeans this issue relates to the parser.
B-indicates an issue which is a Blocker.
E-indicates the Experience required to work on the ticket. E-easy suggests that this issue is a good entry-level ticket which should be easy enough to fix. E-hard is a more involved, more difficult task.
I-indicates the Importance of the issue
P-indicates the Priority of the issue
T-denotes what Team to which the issue belongs.
For beginners, I recommend looking at issues marked with
E-mentor. These issues will be somewhat easy to work on, and have a member of the Rust team who will mentor you through the process, answering questions as they arise.
Most contributions to the Rust compiler will require a corresponding test. There are a few different kinds of tests to be aware of.
NOTE: A more comprehensive guide to the Rust test suite can be found on the rust-wiki-backup
Language & compiler tests
These tests test the compiler against Rust code. There are a few different types of test.
- compile-fail - tests which expect to fail to compile
- run-fail - tests which expect to fail at run time
- parse-fail - tests which expect to fail during parsing
Language & compiler tests can be run with the
make check-stage1 command.
Most crates include unit tests. These tests live inside an inner module called
test. These tests are the same as the unit tests described in the book
Unit tests can be run with the
make check-stage1-crates command.
The build system will also test Rust code snippets from documentation examples.
You can run documentation tests with the
make check-stage1-doc command.
When you have a patch you feel is ready for submission, you should open a pull request against the master branch of the Rust project. This GitHub guide gives a great walk-through of submitting pull requests.
When you submit a new pull request, a reviewer will be automatically appointed by a bot called rust-highfive. Throughout the review process you may be asked to make further modifications to the code. If this happens, you should submit those modifications as new commits (don't amend commits!). This will make it easier for your reviewers to see what you have changed.
When your commit has been approved, a project member will leave a comment instructing the integration bot, called bors, to test & merge your commit. The comment will looks something like:
@bors: r+ <commit-hash>
For more information on pull requests, take a look at the relevant section of the contributing guide
In future posts I will be exploring some of the internal workings of the compiler. If that sorf of thing interests you, check back here soon!