328215 - Litmus testcase <divs> don't display correctly with mac trunk builds

Status: RESOLVED FIXED

(Webtools Graveyard :: Litmus, defect)

Description

[Chris Cooper [:coop] (he/him)]

I have not yet seen this for myself, but Tracy and other have reported that Litmus is unusable on recent Trunk builds and on Safari, i.e. there is no display.

Tracy: can you add more details, and set affected platforms if applicable.

Comment 1

[Tracy Walker [:tracy]]

This is Mac only.
Use Firefox trunk or Safari. 

-Go into a test subgroup to view test cases.
-click on the first (or any) testcase of the subgroup.
tested results:

the testcase opens up as expected. But the content is missing; just a a gray background filling the space. 

Comment 2

[Zach Lipton [:zach]]

This is extremely odd. I can reproduce this with trunk bits on Mac, but not with 1.5. A View Source on the Run Tests page seems to show the source of the _previous_ sysconfig page (perhaps because they both show the same URL?), even though the content is clearly different. Stranger still is that I can select the "missing" content from the empty gray background and paste it into a text editor, revealing the testcast data. Furthermore, if I select a swatch of page and do a View Selection Source, I get the correct page source. It seems that Litmus is sending the correct page (or at least, the same page it sent before), but that it's being mangled in the browser somehow. 

I only have Mac here. Does this happen on any other platform? I'll try to see if I can create a reduced testcase for this. 

Comment 3

[Zach Lipton [:zach]]

Well that was fun. Spent an hour on a reduced testcase, which quickly got me to a magic search string that led me to bug 325296. Basically, trunk builds can't handle having opacities set for text in divs on mac, which is exactly what moo.fx does to create our fancy testcase boxes. It looks like roc and jruderman are looking into a fix, so this should hopefully be fixed soon. In the meantime, we could hack around moo.fx and just hide and show the divs ourselves if needed. 

Comment 4

[Jesse Ruderman]

Is the reduced testcase for Safari the same?

Comment 5

[Zach Lipton [:zach]]

Litmus is broken in Safari for me in a completely different way (though the issue looks similar visually) - the "select subgroup" area never appears on the testgroup selection page. I've filed bug 328268 to track the issue on Safari. 

Unless my detective work is horribly off, this should be fixed when bug 325296 is fixed. 

Comment 6

[Tracy Walker [:tracy]]

Zach and Jesse,

Forgive my inclusion of Safari here.  I remember it being broken and lumped it in.  Thanks to Zach for filing the correct bug on Safari.

Comment 7

[Chris Cooper [:coop] (he/him)]

This sounds like a simple CSS error for me. Can we verify in the litmus.css file that everytime we're setting a background-color that we are also set a font color that will display on that background-color? Even if we rule that out, it's progress.

Comment 8

[Chris Cooper [:coop] (he/him)]

...and turn off the opacity effects until bug 325296 is fixed.

Comment 9

[Zach Lipton [:zach]]

Ok this should be fixed now. I've disabled the opacity effects for Mac users only as a workaround until the gecko bug is fixed. It's working now for me on Mac, but it would be great if someone could test this on both Mac and another platform to make sure I haven't broken anything. 

I've been having some weird issues with the cache on trunk builds where script files never seem to be reloaded, so to pickup the change, you might need to clear your cache first. 

Comment 10

[Chris Cooper [:coop] (he/him)]

I'm Mac-enabled now, so I'll take a look at this.

Comment 11

[Chris Cooper [:coop] (he/him)]

I've verified the fix here on Mac against 1.5(UB) and trunk. Leaving bug open so we can revert once bug 325296 is fixed.

Comment 12

[Chris Cooper [:coop] (he/him)]

Removing blocker status since our workaround is perfectly serviceable for now.

Comment 13

[Chris Cooper [:coop] (he/him)]

The dependency has been fixed (bug 325296)...closing.