40652 2010-06-15 22:18:50 -0700 webkit-patch land generates an incorrect commit message 2011-01-07 15:08:24 -0800 1 1 1 Unclassified WebKit Tools / Tests 528+ (Nightly build) PC OS X 10.5 NEW P2 Normal --- 1 yuzo webkit-unassigned eric mihaip ossy oldest_to_newest 238673 0 yuzo 2010-06-15 22:18:50 -0700 I've committed http://trac.webkit.org/changeset/61232 with webkit-patch land --ignore-builders The commit message was incorrect -- it includes change logs for other changes. More specifically, I did the following: git checkout master git svn rebase git rebase master <mybranch> git reset HEAD^ git add <newfiles> webkit-patch land --ignore-builders Please note that I didn't do resolve-ChangeLogs because git rebase didn't complain of conflicts. This could be a factor. Maybe I made some stupid mistake, but I'd appreciate it if anyone can investigate. 238675 1 yuzo 2010-06-15 22:41:32 -0700 The diff for WebCore/ChangeLog was not from the beginning of the file. That might be the cause? 328646 2 ossy 2011-01-03 10:50:32 -0800 One more fail: http://trac.webkit.org/changeset/74897 Eric, could you check it please? 331038 3 eric 2011-01-07 15:04:21 -0800 webkit-patch land is supposed to warn when ChangeLog diffs start further than 8 lines into the file. 331039 4 eric 2011-01-07 15:05:06 -0800 See http://trac.webkit.org/browser/trunk/Tools/Scripts/webkitpy/tool/steps/validatechangelogs.py#L37 331041 5 mihaip 2011-01-07 15:06:51 -0800 In the case of http://trac.webkit.org/changeset/74897, the problem could've been when I did a "git svn rebase" which auto-merged ChangeLog diffs with resolve-ChangeLogs, which shares no code with webkit-patch. 331044 6 eric 2011-01-07 15:08:24 -0800 Sure, but if you used webkit-patch land to do the commit, then the ValidateChangeLogs step would have run, and that check should have failed, prompting you to check your diff. However it's very possible that either the DiffParser or the logic in that step is wrong.