RESOLVED FIXED 26447
REGRESSION: Animated GIF problems on GTK 
https://bugs.webkit.org/show_bug.cgi?id=26447
Summary REGRESSION: Animated GIF problems on GTK
Peter Kasting
Reported 2009-06-16 10:33:17 PDT
According to bug 25709 comment 73, one of the patches on bug 25709 caused problems with animated GIFs where "at the end of one animation cycle, [we show] an empty (sometimes black) image". According to xan on IRC, there is a problem with GIFs on GTK that "they flicker a lot". I suspect this is the bug zecke is reporting. If so, the problem happened before landing the "part nine" patch on bug 25709. xan reports that the GIF at http://www.cracked.com/funny-36-twilight/ exhibits the problem. I will try and get more info from IRC.
Attachments
patch v1 (14.87 KB, patch)
2009-06-16 11:48 PDT, Peter Kasting 		no flags
DetailsFormatted DiffDiff
patch v1, for real (2.32 KB, patch)
2009-06-16 11:51 PDT, Peter Kasting 		xan.lopez: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Peter Kasting
Comment 1 2009-06-16 10:55:18 PDT
According to IRC trackback zecke reports the problem was caused by r44631 and was indeed in GTK+.
Brent Fulgham
Comment 2 2009-06-16 11:07:20 PDT
I can confirm that the animated GIF "flickers", or shows a black frame under Windows Cairo. I have not payed much attention to animated GIF's, so I cannot confirm when this change might have happened.
Peter Kasting
Comment 3 2009-06-16 11:13:50 PDT
On IRC bfulgham reports that the precise behavior is that the image rapidly cycles between a single still image and a black frame. There is no other discernable animation.
Peter Kasting
Comment 4 2009-06-16 11:48:33 PDT
Created attachment 31359 [details] patch v1 Simple fix. I forgot to copy one member.
Peter Kasting
Comment 5 2009-06-16 11:51:14 PDT
Created attachment 31360 [details] patch v1, for real Oops, uploaded the wrong patch file!
Xan Lopez
Comment 6 2009-06-16 11:55:22 PDT
Comment on attachment 31360 [details] patch v1, for real I can confirm that the missing setting of m_size in the copy function was the cause of the bug (and it looks ok), so r=me
Peter Kasting
Comment 7 2009-06-16 12:27:59 PDT
Fixed in r44733.
Note You need to log in before you can comment on or make changes to this bug.