1210418 - 2x vertical text scaling while scrolling, redux.

Status: RESOLVED WORKSFORME

(Core :: Panning and Zooming, defect)

Description

[Mike Hoye [:mhoye]]

On current nightly I am seeing the behavior described in bug 1195400 again, in particular on medium.com:

https://medium.com/working-parents-in-america/beyond-doing-half-the-parenting-d92996d7a106 

As before: doesn't occur on all sites but is consistent where it does, can reproduce at will, is hideous, E10S enabled.

Comment 1

[Mike Hoye [:mhoye]]

Also as before, Botond's remedy in https://bugzilla.mozilla.org/show_bug.cgi?id=1195400#c3  fixes the issue.

Comment 3

[Markus Stange [:mstange]]

I've determined that at least on https://gumroad.com/l/maintaining-open-source-projects (from duplicate bug 1211272), the problem is bug 1206704 - a big intermediate surface. So relanding bug 1021845 should fix this.

Comment 4

[Liz Henry (:lizzard) (relman/hg->git project)]

Tracking since this is a recent regression. Does it affect other branches too?

Comment 5

[u279076]

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #4)
> Tracking since this is a recent regression. Does it affect other branches
> too?

I'm not seeing the issue locally. Mike, can you please check if this affect Beta and Dev Edition?

Comment 6

[Mike Hoye [:mhoye]]

I am no longer able to duplicate this on current nightly - 44.0a1 (2015-10-13).

Comment 7

[David Anderson [:dvander] - inactive, e-mail if emergency]

I can't reproduce this either. Feel free to re-open if we can get it again.

Comment 8

[Markus Stange [:mstange]]

Apparently it's still reproducible with a non-default zoom level, but it looks like we're using bug 1225868 to track that now.

Comment 9

[David Anderson [:dvander] - inactive, e-mail if emergency]

... actually, I can still reproduce this. But once again only on Mac, with APZ, with tiling, and with bug 1135271 backed out so the mix-blend-mode is accelerated.

I can't reproduce bug 1225868, so I don't yet know if it's similar. This bug looks like a problem in the compositor though. The GL surface looks fine in content and only looks wrong in the compositor's backbuffer.

Comment 10

[Markus Stange [:mstange]]

Can you check whether it still happens in the latest try build in bug 1021845?

Comment 11

[David Anderson [:dvander] - inactive, e-mail if emergency]

Sigh - sorry, I meant to post comment #9 in bug 1184343. Anyway, the try cset in bug 1021845 does appear to fix *that* bug. I can't reproduce this one still.

Comment 12

[Kartikaya Gupta (email:[e-mail])]

Does that mean we can close this one again?

Comment 13

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Updating status-firefox44 as verified based on comment 6.