We can use help in a bunch of areas and any help is appreciated. Our GitHub issues serve as a place for any discussion, whether it's bug reports, questions, project direction etc. As the project grows this policy may change.
Our Discord server is open for help and more adhoc discussion. All activity on the Discord is still moderated and will be strictly enforced under the project's Code of Conduct.
Building this project requires a stable
Rust toolchain, which can be installed using rustup.
Clone the repository and navigate to the tools
directory:
git clone https://github.com/rome/tools
cd tools
Compile all packages and dependencies:
cargo build
Rome can be used via the cli
bin in the rome_cli
package:
cargo run --bin cli -- --help
Rome can be used as a language server by following the instructions below.
The Rome language server is the binary crate rome_lsp
which can be built using:
cargo build --bin rome_lsp
If benchmarking the language server, be sure to build with the --release
flag.
The VS Code extension can be installed from the Marketplace and can be used with a development build of the language server by setting the "rome.lspBin"
VS Code setting to the path of the binary:
"rome.lspBin": "/path/to/rome/target/debug/rome_lsp"
To instead build the VS Code extension from source, navigate to the editors/vscode
directory and run:
npm install
npm run build
This will create a rome_lsp.vsix
which you can install into VS Code by running:
npm run install-extension
The "rome.lspBin"
VS Code setting will still need to be set as described above.
If files specific to your local development environment should be ignored, please add these files to a global git ignore file rather than to a git ignore file within Rome.
You can find more information on this process here.
The Rome website is built with Eleventy. To start a development server you can run the following commands:
cd website
npm install
npm start
cargo lint
is a cargo alias that runs `clippy' - rust official linter - under the hood;cargo format
is a cargo alias that runsrust-fmt
- rust official formatter - under the hood;cargo test
will run the suite; make sure to run this command from the root of the project, so it will run the tests of all the internal crates;
If you work on some parser and you create new nodes or modify existing ones, will need to run a command to update some files that are auto-generated.
This command will update the syntax of the parsers.
The source is generated from the ungram
files.
This command will create new tests for your parser. We currently have a neat infrastructure where tests for parser are generated com inline comments found inside the source code. Please read the proper chapter for more information
It's strongly advised to run this command before committing new changes.
This command will check and report parser conformance against different test suites. We currently target the Official ECMAScript Conformance Test Suite and the Typescript Test Suite
The test suites are included as git submodules and can be pulled using:
git submodule update --init --recursive
Internally, the Rome team adheres as closely as possible to the conventional commit specification. The following this convention encourages commit best-practices and facilitates commit-powered features like change log generation.
The following commit prefixes are supported:
feat:
, a new featurefix:
, a bugfixdocs:
, a documentation updatetest
, a test updatechore:
, project housekeepingperf:
, project performancerefactor:
, refactor of the code without change in functionality
Below are examples of well-formatted commits:
feat(compiler): implement parsing for new type of files
fix: fix nasty unhandled error
docs: fix link to website page
test(lint): add more cases to handle invalid rules
When creating a new pull request, it's preferable to use a conventional commit-formatted title, as this title will be used as the default commit message on the squashed commit after merging.
Here are some other scripts that you might find useful.
If you are a core contributor, and you have access to create new branches from the main repository (not a fork), use these comments to run specific workflows:
!bench_parser
benchmarks the parser's runtime performance and writes a comment with the results;!bench_formatter
benchmarks the formatter runtime performance and writes a comment with the results;
-
Forbid a concept
no<Concept>
When a rule's sole intention is to forbid a single concept - such as disallowing the use of
debugger
statements - the rule should be named using theno
prefix. For example, the rule to disallow the use ofdebugger
statements is namednoDebugger
. -
Mandate a concept
use<Concept>
When a rule's sole intention is to mandate a single concept - such as forcing the use of camel-casing - the rule should be named using the
use
prefix. For example, the rule to mandating the use of camel-cased variable names is nameduseCamelCase
.
To have a better understanding of our parsing infrastructure, please read the in-depth section
If you want to create a new test for an existing parser, you will have to inline the code that you want to test in a comment that is created in a specific way.
Let's say that you created a new parsing feature and you need new tests from scratch, just go to the source code where you parse this new feature if JavaScript, and add the following comment:
// test feature_name
// let a = { new_feature : "" }
// let b = { new_feature : "" }
fn parse_new_feature(p: &mut Parser) -> ParsedSyntax {}
The first line, // test feature_name
the important one. This will tell to the
testing infrastructure to create a positive test (without parsing errors), called
feature_name.js
inside the test_data/inline/ok
folder.
The content of this file will be:
let a = { new_feature : "" }
let b = { new_feature : "" }
Basically, everything after the key comment will be the content of the new file.
Now you need to run cargo codegen test
and the task will actually generate this file for you.
In case you want to create a negative test (with parsing errors), you will create a new comment like this:
// test feature_name
// let a = { new_feature : "" }
// let b = { new_feature : "" }
+ // test_err feature_name
+ // let a = { : "" }
+ // let b = { new_feature : }
fn parse_new_feature(p: &mut Parser) -> ParsedSyntax {}
Mind the different comment test_err
, which marks the error for the test suite
as a test that has to fail.
Run the command cargo codegen test
and you will see a new file called
feature_name.js
inside the test_data/inline/err
folder.
The content of this file will be:
let a = { : "" }
let b = { new_feature : }
Now run the command: Unix/macOS
env UPDATE_EXPECT=1 cargo test
Windows
set UPDATE_EXPECT=1 & cargo test
The command will tell the test suite to generate and update the .rast
files.
If tests that are inside the ok/
folder fail or if tests that are inside the err/
folder don't emit, the whole test suite will fail.
We follow the specs suggested by the official documentation:
Odd minor versions are dedicated to pre-releases, e.g. *.5.*
.
Even minor versions are dedicated to official releases, e.g. *.6.*
.