Skip to content

Latest commit

 

History

History
119 lines (87 loc) · 4.03 KB

CONTRIBUTING.md

File metadata and controls

119 lines (87 loc) · 4.03 KB

Contributing to iai-callgrind

Thank you for your interest in contributing to iai-callgrind!

Feature Requests and Bug reports

Feature requests and bug reports should be reported in the Issue Tracker. Please have a look at existing issues with the enhancement or bug labels.

Patches / Pull Requests

All patches have to be sent on Github as pull requests. Before starting a pull request, it is best to open an issue first so no efforts are wasted.

If you are looking for a place to start contributing to iai-callgrind, take a look at the help wanted or good first issue issues.

The minimum supported version (MSRV) of iai-callgrind is Rust 1.66.0 and all patches are expected to work with the minimum supported version.

All notable changes need to be added to the CHANGELOG.

How to get started

Clone this repo

git clone https://github.com/iai-callgrind/iai-callgrind.git
cd iai-callgrind

Working on this project is a piece of cake with just and if you have the just shell completions installed. Before running any install commands with just, it is recommended to first inspect it with --dry-run. Install the basics needed to start working on this project with:

just install-workspace

This command will install git hooks, the necessary components for the stable, nightly toolchain and the current MSRV toolchain, run some checks for tools which need to be installed, ...

To get an overview over all possible just rules run just -l or directly inspect the Justfile in the root of this project.

If your IDE can handle it, it's usually best to work with the MSRV locally

rustup override set 1.66.0

What is left is to setup your favorite editor to use nightly rustfmt and clippy from the rust stable toolchain in order to pass the formatting and linting checks in the ci.

Testing

iai-callgrind lacks tests and contributions expanding the test suite are also very welcome. Patches have to include tests to verify (at a minimum) that the whole pipeline runs through without errors.

The benches in the benchmark-tests package run the whole pipeline which is good for verifying that there are no panics or errors. If you've made changes in the iai-callgrind-runner package then you can point the IAI_CALLGRIND_RUNNER environment variable to your modified version of the iai-callgrind-runner binary:

cargo build -p iai-callgrind-runner --release
IAI_CALLGRIND_RUNNER=$(readlink -e target/release/iai-callgrind-runner) cargo bench -p benchmark-tests

or with just in a single command:

just bench-test-all

or a specific bench of the benchmark-test package

just bench-test test_lib_bench_tools

The concrete results of the benchmarks are not checked when running the benchmark tests that way. You can use

just full-bench-test test_lib_bench_tools

which also checks that any output files that are expected to be created by the benchmark run are actually there. Depending on the test configuration benchmarks are sometimes run multiple times. Currently, we don't have any tests in place which verify the benchmark terminal outputs. So, for concrete results it's better to use unit tests or integration tests in the iai-callgrind-runner package.

Contact

If there are any outstanding questions about contributing to iai-callgrind, they can be asked on the iai-callgrind issue tracker.