Issue tracker

Mahout's issue tracker is located here. For most changes (apart from trivial stuff) Mahout works according to a review-then-commit model. This means anything that is to be added is first presented as a patch in the issue tracker. All conversations in the issue tracker are automatically echoed on the developer mailing list and people tend to respond or continue conversations there rather in the issue tracker, so in order to follow an issue you also have to read to the mailing lists.

An issue does not literally have to be an issue. It could be a wish, task, bug report, etc. and it does not have to contain a patch.

Mahout uses JIRA by Atlassian.

Best practices

Don't create duplicate issues. Make sure your problem is a problem and that nobody else already fixed it. If you are new to the project, it is often preferred that the subject of an issue is discussed on one of our mailing lists before an issue is created - in particular when it comes to adding new functionality.

Quote only what it is you are responding to in comments.

Patches should be created at trunk or trunk parent level and if possible be a single uncompressed text file so it is easy to inspect the patch in a web browser. (See Patch Check List .)

Use the issue identity when referring to an issue in any discussion. "MAHOUT-n" and not "mahout-n" or "n". MAHOUT-1 would automatically be linked to MAHOUT-1 in a better world.

A note to committers: Make sure to mention the issue id in each commit. Not only has JIRA the capability of auto-linking commits to the issue they are related to that way, it also makes it easier to get further information for a specific commit when browsing through the commit log and within the commit mailing list.