Created attachment 551279 [details]
example GIF affected by this issue
Some GIF images drop all frames after a certain point, causing them to restart prematurely.
The regression range looks to be:
From that list, the likely culprit is Bug 609499
EXPECTED RESULTS: Chair spins, camera pans up/left to a woman's face.
ACTUAL RESULTS: Chair only spins partway, & animation abruptly restarts.
Firefox 5 shows EXPECTED RESULTS. Nightly shows ACTUAL RESULTS. (Nightly never gets to the part of the animation with the woman's face)
Mozilla/5.0 (X11; Linux i686; rv:8.0a1) Gecko/20110806 Firefox/8.0a1
It's happening on Windows platform (win7 at least) also.
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:8.0a1) Gecko/20110806 Firefox/8.0a1
*** Bug 678746 has been marked as a duplicate of this bug. ***
Local track down using Linux x86_64:
The first bad revision is:
user: Joe Drew <firstname.lastname@example.org>
date: Wed Aug 03 15:19:19 2011 -0400
summary: Bug 609499 - Some invalid GIF files don't terminate their LZW data with a 0-length block. Handle them better by handling them less specially. r=jrmuizel
Created attachment 553375 [details]
testcase 2 (should show 2 people colliding)
Here's another affected gif. Should show 2 people colliding, but Nightly just shows just a few frames of one person running at the beginning.
(source: http://www.reddit.com/r/funny/comments/jjr70/smoosh/ )
Created attachment 553725 [details]
testcase, shall rotate around z axis, does not rotate
*** Bug 680437 has been marked as a duplicate of this bug. ***
*** Bug 680775 has been marked as a duplicate of this bug. ***
Another affected gif:http://i.imgur.com/Yg5za.gif
*** Bug 682223 has been marked as a duplicate of this bug. ***
joe, jeff, could you please take a look at this? We need to figure out whether we fix this or back out bug 609499 on Aurora, imo.
I've backed out 67044ea27b53 from FF8 (Aurora) and FF9 (Central):
(In reply to Jeff Muizelaar [:jrmuizel] from comment #12)
> I've backed out 67044ea27b53 from FF8 (Aurora) and FF9 (Central):
That's fixed it, it seems
Why is this bug still open?
I think it can be marked as FIXED or WORKSFORME.
Since bug 609499 was backed out, this is no longer an issue.
*** Bug 679587 has been marked as a duplicate of this bug. ***
qa+ for verification using the attached testcase in comment 0.