Information about the ORC project that is most important for developers working on the project.
The ORC format specification defines the format to promote compatibility between implementations.
We have committers from many different companies. The full list of ORC committers is available.
The most important communication mechanism for the project are its mailing lists. The mailing lists have the advantage that they are publicly archived and work well asynchronously across timezones.
Beside the user mailing list, there are several development mailing lists for ORC:
- email@example.com - Development discussions with archive here
- firstname.lastname@example.org - Bug tracking with archive here
- email@example.com - Git tracking with archive here
You can subscribe to the lists by sending email to firstname.lastname@example.org and unsubscribe by sending email to email@example.com.
Each code change requires a jira to track the discussion of the change.
ORC uses git for version control. Get the source code and configure it to fetch the pull requests also:
% git clone -o apache firstname.lastname@example.org:apache/orc.git % cd orc % git config --add remote.apache.fetch '+refs/pull/*/head:refs/remotes/apache/pr/*'
Pull requests will be named “apache/pr/999” for pull request 999.
If you are a committer, you need to register your Github id with Apache:
- Visit id.apache.org and add your Github id.
- Visit Apache gitbox and get authorization for the projects that use Apache gitbox, which includes ORC.
Once this is done, your Github account will be placed in the ORC github team, so you’ll be able to close PRs, etc.
The important branches are:
- main - The main branch for all development
- branch-X.Y - The release branches
- asf-site - The generated html pages that are deployed as https://orc.apache.org/
Releases are tagged as “rel/release-X.Y.Z”. Apache’s git repository guarantees that tags in the “rel/*” namespace are never deleted or changed.
Please check our coding guidelines.
We’ve added several shortcuts to various relevant pages:
From our website, you can use:
- /bugs to jump to our bug database
- /downloads to jump to our downloads page
- /releases to jump to our releases page
- /src to jump to our source code
- /web-src to jump to our site source code
ORC uses Commit-Then-Review, so patches can be committed without a committer’s review. However, most changes should be reviewed first.
Approving a pull request
Fetch the current state of the project:
% git fetch apache
Switch to the branch:
% git checkout apache/pr/999
You’ll want to rebase it and make it a single commit by squashing the commits into a single commit.
% git rebase -i apache/main
Update the commit message to sign it using your GPG key and close the pull request:
% git commit --amend -s -S
Ensure the first line of the commit starts with the jira number (eg. ORC-123) and includes a description of what was changed. Also add a line such as “Fixes #999”, which asks the Apache infrastructure to close pull request 999. If you wish you close a pull request without claiming to have fixed the problem, the form “Closes #999” also works.
Finally, push the result to Apache:
% git push apache HEAD:main
Creating a GPG key
When you become a committer, you should create a 4096 bit GPG key.
% gpg --full-gen-key
Use 4096 bits and your Apache email address. Once it is created, you’ll need to get your key fingerprint. Avoid using the short fingerprint (eg. 3D0C92B9), because it is possible to generate fake keys that have the same short fingerprint as the real key.
% gpg --list-secret-keys --keyid-format LONG
Your key fingerprint is the string after “rsa4096/”. Example output for the key with fingerprint 1209E7F13D0C92B9 looks like:
/Users/owen/.gnupg/pubring.gpg ------------------------------ sec rsa4096/1209E7F13D0C92B9 2010-02-23 [SC] 47660BC98BC433F01E5C90581209E7F13D0C92B9 uid [ultimate] Owen O'Malley <email@example.com>
Now publish your public key to one of the public keyservers. I usually use hkp://pgp.mit.edu, although any of them will work.
gpg --send-key <your key fingerprint>
Next, you need to update the Apache account database with your new key. Login to add your new key’s fingerprint and your github id. It is also good to update your github profile with your GPG public key as well at https://github.com/settings/keys.
After you’ve created your key, it is good to get someone in the ORC community to sign it for you. Contact someone directly or send email to firstname.lastname@example.org asking for someone to sign it.
Making a Release
The release process for ORC is driven by a Release Manager. They should discuss their intention to start the process on the dev list and then follow the steps of how to release ORC.
Apache expects the projects to manage their current release artifact distribution using subversion. It should be limited to the latest release in each of the active release branches.
The ORC dist directory is managed via svn in https://dist.apache.org/repos/dist/release/orc. The release artifacts are pushed to many mirrors. Files in the dist directory are available forever via the Apache dist archive.
ORC has a set of bylaws that describe the rules for the different votes within our project.