Skip to content

Latest commit

 

History

History
379 lines (286 loc) · 16.6 KB

CONTRIBUTING.md

File metadata and controls

379 lines (286 loc) · 16.6 KB
Raw

Contributing to Cabal

Building Cabal for hacking

If you use the latest version of cabal published on Hackage, it is sufficient to run:

cabal build cabal

If not, you aren't able to build the testsuite, so you need to disable the default cabal.project that implies configuring the testsuite, e.g., with:

cabal build --project-file=cabal.project.release cabal

Note If you're using Nix, you might find it convenient to work within a shell that has all the Cabal development dependencies:

$ nix-shell -p cabal-install ghc ghcid haskellPackages.fourmolu_0_12_0_0 pkgconfig zlib.dev

A Nix flake developer shell with these dependencies is also available, supported solely by the community, through the command nix develop github:yvan-sraka/cabal.nix.

The location of your build products will vary depending on which version of cabal-install you use to build; see the documentation section Where are my build products? to find the binary (or just run find -type f -executable -name cabal).

Here are some other useful variations on the commands:

cabal build Cabal # build library only
cabal build Cabal-tests:unit-tests # build Cabal's unit test suite
cabal build cabal-tests # etc...

Running tests

Using GitHub Actions. If you are not in a hurry, the most convenient way to run tests on Cabal is to make a branch on GitHub and then open a pull request; our continuous integration service on GitHub Actions builds and tests your code. Title your PR with WIP so we know that it does not need code review.

Some tips for using GitHub Actions effectively:

  • GitHub Actions builds take a long time. Use them when you are pretty sure everything is OK; otherwise, try to run relevant tests locally first.

  • If you are only changing documentation in the docs/ subdirectory, or if you change README.md or CONTRIBUTING.md, then we only run a small subset of the CI jobs. You can therefore open small PRs with improvements to the documentation without feeling guilty about wasted resources!

  • Watch over your jobs on the GitHub Actions website. If you know a build of yours is going to fail (because one job has already failed), be nice to others and cancel the rest of the jobs, so that other commits on the build queue can be processed.

How to debug a failing CI test. One of the annoying things about running tests on CI is when they fail, there is often no easy way to further troubleshoot the broken build. Here are some guidelines for debugging continuous integration failures:

  1. Can you tell what the problem is by looking at the logs? The cabal-testsuite tests run with -v logging by default, which is dumped to the log upon failure; you may be able to figure out what the problem is directly this way.

  2. Can you reproduce the problem by running the test locally? See the next section for how to run the various test suites on your local machine.

  3. Is the test failing only for a specific version of GHC, or a specific operating system? If so, try reproducing the problem on the specific configuration.

  4. Is the test failing on a GitHub Actions per-GHC build. In this case, if you click on "Branch", you can get access to the precise binaries that were built by GitHub Actions that are being tested. If you have an Ubuntu system, you can download the binaries and run them directly.

If none of these let you reproduce, there might be some race condition or continuous integration breakage; please file a bug.

Running tests locally. To run tests locally with cabal, you will need to know the name of the test suite you want. Cabal and cabal-install have several. Also, you'll want to read Where are my build products?

The most important test suite is cabal-testsuite: most user-visible changes to Cabal should come with a test in this framework. See cabal-testsuite/README.md for more information about how to run tests and write new ones. Quick start: use cabal-tests to run Cabal tests, and cabal-tests --with-cabal=/path/to/cabal to run cabal-install tests (don't forget --with-cabal! Your cabal-install tests won't run without it).

There are also other test suites:

  • Cabal-tests:unit-tests are small, quick-running unit tests on small pieces of functionality in Cabal. If you are working on some utility functions in the Cabal library you should run this test suite.

  • cabal-install:unit-tests are small, quick-running unit tests on small pieces of functionality in cabal-install. If you are working on some utility functions in cabal-install you should run this test suite.

  • cabal-install:long-tests are QuickCheck tests on cabal-install's dependency solver, VCS, and file monitoring code. If you are working on the solver you should run this test suite.

  • cabal-install:integration-tests2 are integration tests on some top-level API functions inside the cabal-install source code.

For these test executables, -p which applies a regex filter to the test names. When running cabal-install test suites, one need only use cabal test or cabal run <test-target> in order to test locally.

QA Notes

Manual Quality Assurance (QA) is performed to ensure that the changes impacting the command-line interface, whether adding or modifying a behaviour, are tested before being released. This allows us to catch UX regressions and put a human perspective into testing.

Contributions that touch cabal-install are expected to include notes for the QA team. They are a description of an expected result upon calling cabal-install with certain parameters, and should be written in the body of the ticket or PR under their own heading, like this:

For instance:

## QA Notes

Calling cabal haddock-project should produce documentation for the whole cabal project with the following defaults enabled:

  • Documentation lives in ./haddocks
  • The file ./haddocks/index.html should exist

Manual QA is not expected to find every possible bug, but to really challenge the assumptions of the contributor, and to verify that their own testing of their patch is not influenced by their setup or implicit knowledge of the system.

Code Style

We use automated formatting with Fourmolu to enforce a unified style across the code bases. It is checked in the CI process. After installing Fourmolu 0.12, there are some makefile targets to help formatting the code base.

  • make style - Format the Cabal, Cabal-syntax and cabal-install directories.
  • make style-modified - Format files modified in the current tree.
  • make style-commit COMMIT=<ref> - Format files modified between HEAD and the given reference.

Other Conventions

  • Format your commit messages in the standard way.

  • A lot of Cabal does not have top-level comments. We are trying to fix this. If you add new top-level definitions, please Haddock them; and if you spend some time understanding what a function does, help us out and add a comment. We'll try to remind you during code review.

  • If you do something tricky or non-obvious, add a comment.

  • For local imports (Cabal module importing Cabal module), import lists are NOT required (although you may use them at your discretion.) For third-party and standard library imports, please use either qualified imports or explicit import lists.

  • You can use basically any GHC extension supported by a GHC in our support window, except Template Haskell, which would cause bootstrapping problems in the GHC compilation process.

  • Our GHC support window is five years for the Cabal library and three years for cabal-install: that is, the Cabal library must be buildable out-of-the-box with the dependencies that shipped with GHC for at least five years. GitHub Actions checks this, so most developers submit a PR to see if their code works on all these versions of GHC. cabal-install must also be buildable on all supported GHCs, although it does not have to be buildable out-of-the-box. Instead, the cabal-install/bootstrap.sh script must be able to download and install all of the dependencies (this is also checked by CI). Also, self-upgrade to the latest version (i.e. cabal install cabal-install) must work with all versions of cabal-install released during the last three years.

  • Cabal has its own Prelude, in Distribution.Compat.Prelude, that provides a compatibility layer and exports some commonly used additional functions. Use it in all new modules.

  • As far as possible, please do not use CPP. If you must use it, try to put it in a Compat module, and minimize the amount of code that is enclosed by CPP. For example, prefer:

    f :: Int -> Int
#ifdef mingw32_HOST_OS
f = (+1)
#else
f = (+2)
#endif

    over:

    #ifdef mingw32_HOST_OS
f :: Int -> Int
f = (+1)
#else
f :: Int -> Int
f = (+2)
#endif

GitHub Ticket Conventions

Each major Cabal/cabal-install release (e.g. 3.4, 3.6, etc.) has a corresponding GitHub Project and milestone. A ticket is included in a release's project if the release managers are tentatively planning on including a fix for the ticket in the release, i.e. if they are actively seeking someone to work on the ticket.

By contrast, a ticket is milestoned to a given release if we are open to accepting a fix in that release, i.e. we would very much appreciate someone working on it, but are not committing to actively sourcing someone to work on it.

GitHub Pull Request Conventions

Every (non-backport) pull request has to go through a review and get 2 approvals. After this is done, the author of the pull request is expected to add any final touches they deem important and put the merge me label on the pull request. If the author lacks permissions to apply labels, they are welcome to explicitly signal the merge intent on the discussion thread of the pull request, at which point others (e.g., reviewers) apply the label. Merge buttons are reserved for exceptional situations, e.g., CI fixes being iterated on or backports/patches that need to be expedited for a release.

Currently there is a 2 day buffer for potential extra feedback between the last update of a pull request (e.g. a commit, a rebase, an addition of the merge me label) and the moment the Mergify bot picks up the pull request for a merge.

If your pull request consists of several commits, consider using squash+merge me instead of merge me: the Mergify bot will squash all the commits into one and concatenate the commit messages of the commits before merging.

There is also a merge+no rebase label. Use this very sparingly, as not rebasing severely complicates Git history. It is intended for special circumstances, as when the PR branch cannot or should not be modified. If you have any questions about it, please ask us.

Changelog

When opening a pull request with a user-visible change, you should write one changelog entry (or more in case of multiple independent changes) — the information will end up in our release notes.

Changelogs for the next release are stored in the changelog.d directory. The files follow a simple key-value format similar to the one for .cabal files. Free-form text fields (synopsis and description) allow Markdown markup — please, use markup to make our release notes more readable.

Here's an example:

synopsis: Add feature xyz
packages: cabal-install
prs: #0000
issues: #0000 #0000
significance: significant

description: {

- Detail number 1
- Detail number 2

}

Only the synopsis field is actually required, but you should also set the others where applicable.

Field Description
synopsis Brief description of the change. Often just the pr title.
description Longer description, with a list of sub-changes. Not needed for small/atomic changes.
packages Packages affected by the change (cabal-install, Cabal...). Omit if it's an overarching or non-package change.
prs Space-separated hash-prefixed pull request numbers containing the change (usually just one).
issues Space-separated hash-prefixed issue numbers that the change fixes/closes/affects.
significance Set to significant if the change is significant, that is if it warrants being put near the top of the changelog.

You can find a large number of real-world examples of changelog files here.

At release time, the entries will be merged with this tool.

In addition, if you're changing the .cabal file format specification you should add an entry in doc/file-format-changelog.rst.

Communicating

There are a few main venues of communication:

Releases

Notes for how to make a release are at the wiki page "Making a release". Currently, @emilypi, @fgaz and @Mikolaj have access to haskell.org/cabal, and @Mikolaj is the point of contact for getting permissions.

API Documentation

Auto-generated API documentation for the master branch of Cabal is automatically uploaded here: http://haskell.github.io/cabal-website/doc/html/Cabal/.

Issue triage Open Source Helpers

You can contribute by triaging issues which may include reproducing bug reports or asking for vital information, such as version numbers or reproduction instructions. If you would like to start triaging issues, one easy way to get started is to subscribe to cabal on CodeTriage.

Hackage Revisions

We are reactive rather than proactive with revising bounds on our dependencies for code already released on Hackage. If you would benefit from a version bump, please, open a ticket and get familiar with our revision policy.

The burden of proof that the bump is harmless remains with you, but we have a CI setup to show that our main pipeline ("Validate") is fine with the bump. To use it, someone with enough permissions needs to go on the Validate workflow page and dispatch it manually by clicking "Run workflow".

Running workflow manually as discussed above requires you to supply two inputs:

allow-newer line constraints line

Going via an example, imagine that Cabal only allows tar or version less then or equal to 0.6, and you want to bump it to 0.6. Then, to show that Validate succeeds with tar 0.6, you should input

  • tar to the "allow-newer line"
  • tar ==0.6 to the "constraints line"

Hopefully, running the Validate pipeline with these inputs succeeds and you supply the link to the run in the ticket about bumping the bound and making a revision.

If interested in technical details, refer to the parts of validate.yml that mention hackage-revisions.