Bug 86786 – mathml/presentation/mo-stretch.html and mroot-pref-width.html tests fail on Mac

• Component: MathML
• Status: RESOLVED FIXED

[Brady Eidson]

Two mathml/presentation tests fail on Mac.

mathml/presentation/mo-stretch.html
mathml/presentation/mroot-pref-width.html

http://build.webkit.org/results/Lion%20Debug%20(Tests)/r117482%20(6690)/results.html includes an example of the failures.

Due to gaps in testing coverage, it's unclear when these failures were introduced.
They started failing sometime between 117198 and 117355

[Brady Eidson]

Skipped in http://trac.webkit.org/changeset/117514

[Radar WebKit Bug Importer]

<rdar://problem/11479310>

[Dave Barton]

It looks like it may be from r117339. That patch is relevant (tables), and seems to have caused a lot of rebaselining - see http://trac.webkit.org/search?q=117339

I will look at this after my current patch, if no one else does.

[Dave Barton]

The two mathml tests that r117339 made fail actually are:

mathml/presentation/row-alignment.xhtml
mathml/presentation/tables.xhtml

These are already in LayoutTests/platform/mac-lion/Skipped, along with all the other mathml pixel tests, except the two you're skipping here. Your two (mo-stretch.html and mroot-pref-width.html) are newer tests that I added fairly recently, which is why they weren't already skipped.

I have a fix for row-alignment.xhtml and tables.xhtml that I will submit shortly. I would also like to get all these tests passing, or find out why not. The mathml tests do all pass for me on Snow Leopard. The link you give showing your failures is now broken, and I'm running on Snow Leopard, so I can't verify things on Lion. What is the best way for me to proceed? Should I attach a patch that removes some Skipped lines, and see what the Lion bot(s) do with these tests now? Will I be able to see and/or download the output on Lion so I can troubleshoot it when necessary?

Sorry, I am fairly new to WebKit (just made a committer). Thanks in advance for any suggestions or instructions.

[Dave Barton]

Created attachment 144121 [details]
Patch

[Dave Barton]

r117339 actually also made mo-stretch.html fail, but I don't know about mroot-pref-width.html. I've uploaded a patch to hopefully see.

[Dave Barton]

Thanks for the review+! Is there a way to test this patch on e.g. the Lion Debug (Tests) bot other than committing it? I've yet to land my first patch, let alone back one out, so I'm looking for some guidance or help here, thanks!

[Julien Chaffraix]

(In reply to comment #7)
> Thanks for the review+! Is there a way to test this patch on e.g. the Lion Debug (Tests) bot other than committing it? I've yet to land my first patch, let alone back one out, so I'm looking for some guidance or help here, thanks!

No, only the Chromium Linux EWS runs the tests at the moment so your only choice is to commit the patch and see what happens. You can always preemptively disable the tests that you know will require some new baseline to avoid turning the bots red. Also it's better to be on IRC around the time you land to coordinate with the gardener / bot watcher.

[Dave Barton]

Committed r118713: <http://trac.webkit.org/changeset/118713>

[Dave Barton]

Thanks Julien for all the info. I landed the patch, and mo-stretch.html failed, so I then re-skipped these two tests pending further investigation.

Manually committed r118717: http://trac.webkit.org/changeset/118717

[Dave Barton]

Committed r118935: <http://trac.webkit.org/changeset/118935>

[Dave Barton]

Manually committed r118939: http://trac.webkit.org/changeset/118939

I am rebaselining and unskipping mathml tests. My understanding is this doesn't need to be reviewed, but if I'm wrong please let me know, thanks!

[Dave Barton]

The mathml/presentation tests all ran and passed:

http://build.webkit.org/results/Lion%20Release%20%28Tests%29/r118940%20%288805%29/tests_run0.txt
http://build.webkit.org/results/Lion%20Release%20%28Tests%29/r118940%20%288805%29/