In cases where a test started failing but the regression range is non-empty (i.e., there are multiple possibly-guilty revisions), it would be nice for TestFailures to look at the commit logs for the possibly-guilty revisions and see if any of them modified the affected test. Then it could call out those revisions as particularly suspicious.
We can get commit logs from Trac, but we'll need to add some CORS headers to trac.webkit.org to allow XHRs from build.webkit.org to succeed.
Created attachment 99355 [details] Identify revisions that modified failing tests on TestFailures page
Comment on attachment 99355 [details] Identify revisions that modified failing tests on TestFailures page Note that this patch won't actually work until we add CORS headers to trac.webkit.org to allow XHRs from build.webkit.org to work.
Comment on attachment 99355 [details] Identify revisions that modified failing tests on TestFailures page r=me
Committed r90153: <http://trac.webkit.org/changeset/90153>