I've been noticing this since autumn. It used to work properly before. I have this in a stylish script: #bookmarksPanel, #history-panel, #sidebar, #sidebar-header, #sidebar-box { /*background-color: transparent !important;*/ background-color: rgba(255,255,255,0.2) !important; } I only change the background color, but for some reason the text gets semi-transparent, by the looks of it. Beta and stable seem to be unaffected so far.

Previously I reported it to the stylish dev. But it's apparently not a stylish bug: https://github.com/JasonBarnabe/stylish/issues/210

> I've been noticing this since autumn. Would you be willing to bisect on nightlies to see when the problem appeared for you?

If I must. Do nightlies even go back that far?

Very much so, yes. The tinderbox on-checkin builds go back a few months; we have nightlies for years back. I'd offer to do it myself, but there's really not enough information in comment 0 for me to be able to reproduce the bug. :(

Okay, so I tried nightlies: Last good: ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2014/10/2014-10-01-03-02-05-mozilla-central/ First bad: ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2014/10/2014-10-02-03-02-02-mozilla-central/ Specifically I tried these builds: ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2014/10/2014-10-01-03-02-05-mozilla-central/firefox-35.0a1.en-US.win64-x86_64.installer.exe ftp://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/2014/10/2014-10-02-03-02-02-mozilla-central/firefox-35.0a1.en-US.win64-x86_64.installer.exe

Thank you for doing that! That first build is from rev 14665b1de5ee, the second is from rev 2399d1ae89e9, so the changes between those nightlies are: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=14665b1de5ee&tochange=2399d1ae89e9 The most likely thing in there is bug 902952. Could you try setting the "gfx.direct2d.use1_1" preference to false in about:config and changing the "gfx.content.azure.backends" preference to "direct2d,cairo" to see if that's the problem? Might require a restart...

It appears to fix the issue.

[Tracking Requested - why for this release]: Rendering regression Great, thanks! Bas, can you please take a look at this?

[Tracking Requested - why for this release]:

We disabled D2D 1.1 in 36 in beta 9. I don't think the current beta is affected anymore. However, shipping D2D 1.1 in 37 is part of our objective for this release. Tracking it for 37.

(In reply to Boris Zbarsky [:bz] from comment #10) > [Tracking Requested - why for this release]: I wonder if this is another layout bug with layout incorrectly determining the transparency of the text background as in bug 1102896.

This is not actionable until bug 1102896 is fixed? Once it is, is there anything left in this bug?

The tree body item doesn't report any component alpha bounds but I guess it draws any containing text with default settings (it doesn't seem to disable it anywhere) and that means subpixel AA. I just made it report it's whole bounds a component alpha, which is a little overkill, we could do much better but that would require looking at the tree structure pieces. And we also have to support disabled component alpha.

Bug 1102896 doesn't seem to be related.

Uplift perhaps? :)

Comment on attachment 8568831 [details] [diff] [review] patch Approval Request Comment [Feature/regressing bug #]: bug 902952 [User impact if declined]: text in xul tree elements not over opaque pixels will like translucent or disappear [Describe test coverage new/current, TreeHerder]: nope [Risks and why]: should be pretty low risk [String/UUID change made/needed]: none

(In reply to Timothy Nikkel (:tn) from comment #19) > [Describe test coverage new/current, TreeHerder]: nope Given that we have no test coverage, can you please verify on m-c before uplift?

I checked on my machine on m-c and it was fixed, but that doesn't really add much new information since I also checked that it fixed the bug when I made the patch.

(In reply to Timothy Nikkel (:tn) from comment #21) > I checked on my machine on m-c and it was fixed, but that doesn't really add > much new information since I also checked that it fixed the bug when I made > the patch. Thanks Timothy. Avada - As the reporter, if you have time, can you please confirm that you can no longer reproduce this issue in the latest Nightly build?

Comment on attachment 8568831 [details] [diff] [review] patch Tested on m-c. Let's get this into Beta 2. Beta+ Aurora+

(In reply to Lawrence Mandel [:lmandel] (use needinfo) from comment #22) > (In reply to Timothy Nikkel (:tn) from comment #21) > > I checked on my machine on m-c and it was fixed, but that doesn't really add > > much new information since I also checked that it fixed the bug when I made > > the patch. > > Thanks Timothy. > > Avada - As the reporter, if you have time, can you please confirm that you > can no longer reproduce this issue in the latest Nightly build? It looks okay now with d2d 1.1 on.

Thanks Avada. Marking 39 as verified.