Contributing to Apache Arrow¶
Thanks for your interest in the Apache Arrow project. Arrow is a large project and may seem overwhelming when you’re first getting involved. Contributing code is great, but that’s probably not the first place to start. There are lots of ways to make valuable contributions to the project and community.
This page provides some orientation for how to get involved. It also offers some recommendations on how to get best results when engaging with the community.
Code of Conduct¶
All participation in the Apache Arrow project is governed by the ASF’s Code of Conduct.
Join the mailing lists¶
A good first step to getting involved in the Arrow project is to join the mailing lists and participate in discussions where you can. Projects in The Apache Software Foundation (“the ASF”) use public, archived mailing lists to create a public record of each project’s development activities and decision-making process. While lacking the immediacy of chat or other forms of communication, the mailing lists give participants the opportunity to slow down and be thoughtful in their responses, and they help developers who are spread across many timezones to participate more equally.
See the community page for links to subscribe to the mailing lists and to view archives.
Report bugs and propose features¶
Using the software and sharing your experience is a very helpful contribution itself. Those who actively develop Arrow need feedback from users on what works and what doesn’t. Alerting us to unexpected behavior and missing features, even if you can’t solve the problems yourself, help us understand and prioritize work to improve the libraries.
We use JIRA to manage our development “todo” list and to maintain changelogs for releases. In addition, the project’s Confluence site has some useful higher-level views of the JIRA issues.
To create a JIRA issue, you’ll need to have an account on the ASF JIRA, which you can sign yourself up for. The JIRA server hosts bugs and issues for multiple Apache projects. The JIRA project name for Arrow is “ARROW”.
You don’t need any special permissions on JIRA to be able to create issues. Once you are more involved in the project and want to do more on JIRA, such as assign yourself an issue, you will need “Contributor” permissions on the Apache Arrow JIRA. To get this role, ask on the mailing list for a project maintainer’s help.
Tips for using JIRA¶
Before you create a new issue, we recommend you first search among existing Arrow issues.
When reporting a new issue, follow these conventions to help make sure the right people see it:
Use the Component field to indicate the area of the project that your issue pertains to (for example “Python” or “C++”).
Also prefix the issue title with the component name in brackets, for example
[Python] issue name; this helps when navigating lists of open issues, and it also makes our changelogs more readable.If you’re reporting something that used to work in a previous version but doesn’t work in the current release, you can add the “Affects version” field. For feature requests and other proposals, “Affects version” isn’t appropriate.
Project maintainers may later tweak formatting and labels to help improve their visibility. They may add a “Fix version” to indicate that they’re considering it for inclusion in the next release, though adding that tag is not a commitment that it will be done in the next release.
Tips for successful bug reports¶
No one likes having bugs in their software, and in an ideal world, all bugs would get fixed as soon as they were reported. However, time and attention are finite, especially in an open-source project where most contributors are participating in their spare time. All contributors in Apache projects are volunteers and act as individuals, even if they are contributing to the project as part of their job responsibilities.
In order for your bug to get prompt attention, there are things you can do to make it easier for contributors to reproduce and fix it. When you’re reporting a bug, please help us understand the issue by providing, to the best of your ability,
Clear, minimal steps to reproduce the issue, with as few non-Arrow dependencies as possible. If there’s a problem on reading a file, try to provide as small of an example file as possible, or code to create one. If your bug report says “it crashes trying to read my file, but I can’t share it with you,” it’s really hard for us to debug.
Any relevant operating system, language, and library version information
If it isn’t obvious, clearly state the expected behavior and what actually happened.
If a developer can’t get a failing unit test, they won’t be able to know that the issue has been identified, and they won’t know when it has been fixed. Try to anticipate the questions you might be asked by someone working to understand the issue and provide those supporting details up front.
Other resources:
Improve documentation¶
A great way to contribute to the project is to improve documentation. If you found some docs to be incomplete or inaccurate, share your hard-earned knowledge with the rest of the community.
Documentation improvements are also a great way to gain some experience with our submission and review process, discussed below, without requiring a lot of local development environment setup. In fact, many documentation-only changes can be made directly in the GitHub web interface by clicking the “edit” button. This will handle making a fork and a pull request for you.
Contribute code¶
Code contributions, or “patches”, are delivered in the form of GitHub pull requests against the github.com/apache/arrow repository.
Before starting¶
You’ll first need to select a JIRA issue to work on. Perhaps you’re working on one you reported yourself. Otherwise, if you’re looking for something, search the open issues. Anything that’s not in the “In Progress” state is fair game, even if it is “Assigned” to someone, particularly if it has not been recently updated. When in doubt, comment on the issue asking if they mind if you try to put together a pull request; interpret no response to mean that you’re free to proceed.
Please do ask questions, either on the JIRA itself or on the dev mailing list, if you have doubts about where to begin or what approach to take. This is particularly a good idea if this is your first code contribution, so you can get some sense of what the core developers in this part of the project think a good solution looks like. For best results, ask specific, direct questions, such as:
Do you think $PROPOSED_APPROACH is the right one?
In which file(s) should I be looking to make changes?
Is there anything related in the codebase I can look at to learn?
If you ask these questions and do not get an answer, it is OK to ask again.
Pull request and review¶
To contribute a patch:
Submit the patch as a GitHub pull request against the master branch. For a tutorial, see the GitHub guides on forking a repo and sending a pull request. So that your pull request syncs with the JIRA issue, prefix your pull request name with the JIRA issue id (ex: ARROW-767: [C++] Filesystem abstraction).
Give the pull request a clear, brief description: when the pull request is merged, this will be retained in the extended commit message.
Make sure that your code passes the unit tests. You can find instructions how to run the unit tests for each Arrow component in its respective README file.
Core developers and others with a stake in the part of the project your change affects will review, request changes, and hopefully indicate their approval in the end. To make the review process smooth for everyone, try to
Break your work into small, single-purpose patches if possible. It’s much harder to merge in a large change with a lot of disjoint features, and particularly if you’re new to the project, smaller changes are much easier for maintainers to accept.
Add new unit tests for your code.
Follow the style guides for the part(s) of the project you’re modifying. Some languages (C++, Python, and Rust, for example) run a lint check in continuous integration. For all languages, see their respective developer documentation and READMEs for style guidance. In general, try to make it look as if the codebase has a single author, and emulate any conventions you see, whether or not they are officially documented or checked.
When tests are passing and the pull request has been approved by the interested parties, a committer will merge the pull request. This is done with a command-line utility that does a squash merge, so all of your commits will be registered as a single commit to the master branch; this simplifies the connection between JIRA issues and commits, makes it easier to bisect history to identify where changes were introduced, and helps us be able to cherry-pick individual patches onto a maintenance branch.
A side effect of this way of
merging is that your pull request will appear in the GitHub interface to have
been “closed without merge”. Do not be alarmed: if you look at the bottom, you
will see a message that says @user closed this in $COMMIT. In the commit
message of that commit, the merge tool adds the pull request description, a
link back to the pull request, and attribution to the contributor and any
co-authors.
Local git conventions¶
If you are tracking the Arrow source repository locally, here are some tips
for using git.
All Arrow contributors work off of their personal fork of apache/arrow
and submit pull requests “upstream”. Once you’ve cloned your fork of Arrow,
be sure to:
$ git remote add upstream https://github.com/apache/arrow
to set the “upstream” repository.
You are encouraged to develop on branches, rather than your own “master” branch,
and it helps to keep your fork’s master branch synced with upstream/master.
To start a new branch, pull the latest from upstream first:
$ git fetch upstream
$ git checkout master
$ git pull --ff-only upstream master
$ git checkout -b $BRANCH
It does not matter what you call your branch. Some people like to use the JIRA number as branch name, others use descriptive names.
Once you have a branch going, you should sync with upstream/master
regularly, as many commits are merged to master every day.
It is recommended to use git rebase rather than git merge.
To sync your local copy of a branch, you may do the following:
$ git pull upstream $BRANCH --rebase
This will rebase your local commits on top of the tip of upstream/$BRANCH. In case
there are conflicts, and your local commit history has multiple commits, you may
simplify the conflict resolution process by squashing your local commits into a single
commit. Preserving the commit history isn’t as important because when your
feature branch is merged upstream, a squash happens automatically. If you choose this
route, you can abort the rebase with:
$ git rebase --abort
Following which, the local commits can be squashed interactively by running:
$ git rebase --interactive ORIG_HEAD~n
Where n is the number of commits you have in your local branch. After the squash,
you can try the merge again, and this time conflict resolution should be relatively
straightforward.
If you set the following in your repo’s .git/config, the --rebase option can be
omitted from the git pull command, as it is implied by default.
[pull]
rebase = true
Once you have an updated local copy, you can push to your remote repo. Note, since your remote repo still holds the old history, you would need to do a force push.
$ git push --force origin branch
Note about force pushing to a branch that is being reviewed: if you want reviewers to look at your updates, please ensure you comment on the PR on GitHub as simply force pushing does not trigger a notification in the GitHub user interface.
Also, once you have a pull request up, be sure you pull from origin
before rebasing and force-pushing. Arrow maintainers can push commits directly
to your branch, which they sometimes do to help move a pull request along.
In addition, the GitHub PR “suggestion” feature can also add commits to
your branch, so it is possible that your local copy of your branch is missing
some additions.
Guidance for specific features¶
From time to time the community has discussions on specific types of features and improvements that they expect to support. This section outlines decisions that have been made in this regard.
Endianness¶
The Arrow format allows setting endianness. Due to the popularity of little endian architectures most of implementation assume little endian by default. There has been some effort to support big endian platforms as well. Based on a mailing-list discussion, the requirements for a new platform are:
A robust (non-flaky, returning results in a reasonable time) Continuous Integration setup.
Benchmarks for performance critical parts of the code to demonstrate no regression.
Furthermore, for big-endian support, there are two levels that an implementation can support:
Native endianness (all Arrow communication happens with processes of the same endianness). This includes ancillary functionality such as reading and writing various file formats, such as Parquet.
Cross endian support (implementations will do byte reordering when appropriate for IPC and Flight messages).
The decision on what level to support is based on maintainers’ preferences for complexity and technical risk. In general all implementations should be open to native endianness support (provided the CI and performance requirements are met). Cross endianness support is a question for individual maintainers.
The current implementations aiming for cross endian support are:
C++
Implementations that do not intend to implement cross endian support:
Java
For other libraries, a discussion to gather consensus on the mailing-list should be had before submitting PRs.