wiki:CodeReview

Version 2 (modified by levin@chromium.org, 14 years ago) (diff)

--

Once you put a patch up for review (i.e. mark a patch r?), it goes into the Review Queue. That may not be enough to get it reviewed in a timely manner. WebKit reviewers are all overloaded. The responsibility is on the committer to make it as easy as possible to get the code reviewed. To that end, you should:

  • Look at the people who most recently modified the code you're changing and explicitly CC them on the bug.
  • Add a layout test when you fix a bug/change a behavior or explain why that isn't possible and add a manual test to WebCore/manual-tests.
  • Make your patch as small as possible. Really, seriously, go out of your way to make sure you've broken down your patch into the smallest self-contained chunks possible.
  • Only do/fix one thing per patch. Don't change a behavior and then add a fix for an incidental thing that you noticed while doing the original change. It should be two patches.
  • Only do one patch per bug. Multiple iterations of the same patch is fine but use a new bug for a new patch. Otherwise, it gets confusing to read through all of the comments and understand which ones apply to each patch.
  • Do style cleanup or other significant refactoring in a separate patch, preferably as a precursor to the patch you want to land. Especially end of line whitespace cleanup as this just adds a lot of visual noise to the review.
  • Add short per function comments in the change log (see any of Darin Adler's changes for a good example). These help reviewers quickly understand why you are doing a change in that function.
  • Fix errors from any bots that run after you upload your patch.

If you've done all the above, then you can also ping on #webkit to see if there are any reviewers who can review it for you. If it's a trivial change, than anyone can review it. If it's a complicated or gnarly part of the codebase, you'll likely need to of the approval of one of the people who recently modified the code.