We compute the revision against LayoutTests for layout and performance tests but git.svn_revision fetches the latest svn revision in LayoutTests, unlike svn.svn_revision that retrieves the latest svn revision in the entire repository.
Unfortunately, I don't know git svn well enough to fix this bug.
Maybe we need to ignore the path in git.svn_revision since you can't really have a partial git checkout like we can with svn.
andersca suggests we can walk up the directories and look for .git but I wonder if there's a more efficient way of finding the root of a git module?
I'm sure mithro can help us here once SYD is awake on monday (which is sunday for us).
Created attachment 187023 [details] Fixes the bug
Comment on attachment 187023 [details] Fixes the bug Attachment 187023 [details] did not pass win-ews (win): Output: http://queues.webkit.org/results/16435129
(In reply to comment #6) > (From update of attachment 187023 [details]) > Attachment 187023 [details] did not pass win-ews (win): > Output: http://queues.webkit.org/results/16435129 I've notified rfong about the mysterious wtf/Platform.h not found error.
Comment on attachment 187023 [details] Fixes the bug Clearing flags on attachment: 187023 Committed r142178: <http://trac.webkit.org/changeset/142178>
All reviewed patches have been landed. Closing bug.