Bug 86355 – [EFL][DRT] Fix WebCore library path

• Component: WebKit EFL
• Status: RESOLVED FIXED

[Kangil Han]

Fix WebCore library path used on DRT.

[Kangil Han]

Created attachment 141688 [details]
Patch

[Dirk Pranke]

Comment on attachment 141688 [details]
Patch

looks fine, though I'm not an EFL expert so I'll trust that this is right.

[WebKit Review Bot]

Comment on attachment 141688 [details]
Patch

Clearing flags on attachment: 141688

Committed r116981: <http://trac.webkit.org/changeset/116981>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Chris Dumez]

103 new failures on the build bot now:
http://build.webkit.org/builders/EFL%20Linux%20Debug/builds/1639

I'm not 100% sure this is caused by this patch because there were 2 others in that batch. However, this seems the most likely. Any idea?

[Dirk Pranke]

It looks like those are all compositor failures, and according to http://build.webkit.org/builders/EFL%20Linux%20Debug/builds/1639,

http://trac.webkit.org/changeset/116979 also was a part of that build. It seems more likely to me that the compositing change broke the compositing tests?

[Adrienne Walker]

(In reply to comment #6)
> It looks like those are all compositor failures, and according to http://build.webkit.org/builders/EFL%20Linux%20Debug/builds/1639,
> 
> http://trac.webkit.org/changeset/116979 also was a part of that build. It seems more likely to me that the compositing change broke the compositing tests?

My change is a Chromium-only change that adds asserts, so it seems unlikely that it would cause anything other than crashes on debug Chromium bots.

[Dirk Pranke]

(In reply to comment #7)
> (In reply to comment #6)
> > It looks like those are all compositor failures, and according to http://build.webkit.org/builders/EFL%20Linux%20Debug/builds/1639,
> > 
> > http://trac.webkit.org/changeset/116979 also was a part of that build. It seems more likely to me that the compositing change broke the compositing tests?
> 
> My change is a Chromium-only change that adds asserts, so it seems unlikely that it would cause anything other than crashes on debug Chromium bots.

Okay, that's what I get for not looking. Well, it's easy enough to revert this change and see if things green up.

[WebKit Review Bot]

Re-opened since this is blocked by 86402

[Thiago Marcos P. Santos]

Reopening. I think adding myself as CC to the bug collided with the buildbot changing the status of this bug.

[Raphael Kubo da Costa (:rakuco)]

For posterity: the problem is that since <http://trac.webkit.org/changeset/107020> the shared libraries are being built in a different place in the build directory. However, that change forgot to do what the current patch does. But this also meant webcore_efl.so was never found, so webkitpy always considered compositing, mathml and friends were not enabled and the respective tests were not run, and they were thus not rebased.

tl;dr: Kangil's patch is correct, but he should have rebaselined the failing tests. For now, we have rolled his patch out to keep the bot green.

[Kangil Han]

Created attachment 142473 [details]
Patch

[Kangil Han]

(In reply to comment #11)
> For posterity: the problem is that since <http://trac.webkit.org/changeset/107020> the shared libraries are being built in a different place in the build directory. However, that change forgot to do what the current patch does. But this also meant webcore_efl.so was never found, so webkitpy always considered compositing, mathml and friends were not enabled and the respective tests were not run, and they were thus not rebased.
> 
> tl;dr: Kangil's patch is correct, but he should have rebaselined the failing tests. For now, we have rolled his patch out to keep the bot green.

Done!

[Raphael Kubo da Costa (:rakuco)]

Comment on attachment 142473 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=142473&action=review

Thanks for working on this. We also have PNGs in compositing/ and mathml/ that should be updated. Please run r-w-t with --pixel-tests for these directories. Doing so even helps checking these new results are correct with `webkit-patch rebaseline-server'.

> LayoutTests/platform/efl/test_expectations.txt:-463
> -// This test fails if mixer is not at 100%
> -BUGWK83704 : media/event-attributes.html = TEXT

This is wrong and unrelated to this patch.

[Kangil Han]

Please commit BUG 86956, 86957 first.

[Kangil Han]

Created attachment 143261 [details]
Patch

[WebKit Review Bot]

Attachment 143261 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'LayoutTests/ChangeLog', u'LayoutTests/plat..." exit_code: 1
LayoutTests/platform/efl/test_expectations.txt:119:  fast/events/show-modal-dialog-onblur-onfocus.html is also in a Skipped file.  [test/expectations] [5]
Total errors found: 1 in 4 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Kangil Han]

The error was not come from my patch.
I think it should be fixed in different patch.

[Thiago Marcos P. Santos]

LGTM. Just rebase to make the style checker happy.

It doesn't fail for me here locally.

[Kangil Han]

Created attachment 143552 [details]
Patch

[Raphael Kubo da Costa (:rakuco)]

Comment on attachment 143552 [details]
Patch

Informal r+, thank you all for fixing this!

[WebKit Review Bot]

Comment on attachment 143552 [details]
Patch

Clearing flags on attachment: 143552

Committed r118311: <http://trac.webkit.org/changeset/118311>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.