[Webkit-unassigned] [Bug 88591] Regression: garden-o-matic rebaseline does the wrong thing for missing expectations

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=88591





--- Comment #2 from Ojan Vafai <ojan at chromium.org>  2012-06-07 16:40:47 PST ---
(In reply to comment #1)
> hm. I would have thought we should do nothing for MISSING files, but I suppose I can see the argument for just adding them.

This happens all the time because someone will only add the png for one platform, e.g. http://trac.webkit.org/changeset/119760.

> We don't know whether the missing result is txt, png, or both, though, so I don't know how we can avoid trying to fetch both.

I think that's fine. We can put in a FIXME to properly communicate with result is missing in the future. In the meantime, there will be the possibility that rebaseline will do the wrong thing in rare cases. Almost always, this happens for new tests, so I don't think getting this right is super urgent.

> after the patch I just landed, rebaseline-expectations will also ignore the file. In that case, though, the user could just change 'MISSING' to TEXT/IMAGE/IMAGE+TEXT as appropriate to work around the issue.

I'm OK with that for this case. Would be good for the tool at least to give some feedback that it's skipping this though instead of silently doing nothing.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.