Re: webkit 1.2.7
On Sun, Mar 06, 2011 at 09:57:32PM -0500, Todd Carson wrote: On Sun, Mar 06, 2011 at 07:57:34AM -0600, Marco Peereboom wrote: I can confirm that backing that out fixes the regression. On Sun, Mar 06, 2011 at 09:28:58AM +0100, Landry Breuil wrote: Backing out http://gitorious.org/webkitgtk/stable/commit/d842bb085aceec4fcfc392a7b76c92c258cd5151 fixes the issue. https://bugs.webkit.org/show_bug.cgi?id=55830 It didn't work for me. The gif at www.openbsd.org/art1.html still just flickers. This was on amd64. I recompiled webkit with a couple of other changes in addition to backing out the above diff: - the DNS prefetching patch from another thread was applied - the JavaScript JIT compiler was disabled in order to test an interpreter bug fix If it works for other people, maybe one of those somehow re-introduced the regression for me, I don't know. The fix is wrong, looks like it's a typo in GIFImageDecoder.cpp. I'll fix it with another batch of patches tonight. Landry
Re: webkit 1.2.7
the bug has been fixed upstream thanks to dhill http://gitorious.org/webkitgtk/stable/commit/bd6bf8e203cabe7073d4880c64a1ab86ebc1264b On Tue, Mar 08, 2011 at 03:27:18PM +0100, Landry Breuil wrote: On Sun, Mar 06, 2011 at 09:57:32PM -0500, Todd Carson wrote: On Sun, Mar 06, 2011 at 07:57:34AM -0600, Marco Peereboom wrote: I can confirm that backing that out fixes the regression. On Sun, Mar 06, 2011 at 09:28:58AM +0100, Landry Breuil wrote: Backing out http://gitorious.org/webkitgtk/stable/commit/d842bb085aceec4fcfc392a7b76c92c258cd5151 fixes the issue. https://bugs.webkit.org/show_bug.cgi?id=55830 It didn't work for me. The gif at www.openbsd.org/art1.html still just flickers. This was on amd64. I recompiled webkit with a couple of other changes in addition to backing out the above diff: - the DNS prefetching patch from another thread was applied - the JavaScript JIT compiler was disabled in order to test an interpreter bug fix If it works for other people, maybe one of those somehow re-introduced the regression for me, I don't know. The fix is wrong, looks like it's a typo in GIFImageDecoder.cpp. I'll fix it with another batch of patches tonight. Landry
Re: webkit 1.2.7
I can confirm that backing that out fixes the regression. On Sun, Mar 06, 2011 at 09:28:58AM +0100, Landry Breuil wrote: On Sat, Mar 05, 2011 at 06:14:53PM -0500, David Hill wrote: On Fri, Mar 04, 2011 at 11:00:00PM +0100, Landry Breuil wrote: :On Fri, Mar 04, 2011 at 03:48:51PM -0600, Marco Peereboom wrote: : It is definitively a regression. A bad one too. Maybe an interaction : with the lib that renders the gifs? : :The only thing i see is https://bugs.webkit.org/show_bug.cgi?id=46437, :which touchs the Gif backend. See if reverting :http://trac.webkit.org/changeset/68446/trunk/WebCore/platform/image-decoders/gif/GIFImageDecoder.cpp :helps.. but i don't think it's the way to go, as it's supposed to fix a :crasher. Backing out http://gitorious.org/webkitgtk/stable/commit/d842bb085aceec4fcfc392a7b76c92c258cd5151 fixes the issue. https://bugs.webkit.org/show_bug.cgi?id=55830 That looks very strange.. i can't test it atm, can someone else confirm it ? Landry
Re: webkit 1.2.7
On Sun, Mar 06, 2011 at 07:57:34AM -0600, Marco Peereboom wrote: I can confirm that backing that out fixes the regression. On Sun, Mar 06, 2011 at 09:28:58AM +0100, Landry Breuil wrote: Backing out http://gitorious.org/webkitgtk/stable/commit/d842bb085aceec4fcfc392a7b76c92c258cd5151 fixes the issue. https://bugs.webkit.org/show_bug.cgi?id=55830 It didn't work for me. The gif at www.openbsd.org/art1.html still just flickers. This was on amd64. I recompiled webkit with a couple of other changes in addition to backing out the above diff: - the DNS prefetching patch from another thread was applied - the JavaScript JIT compiler was disabled in order to test an interpreter bug fix If it works for other people, maybe one of those somehow re-introduced the regression for me, I don't know.
Re: webkit 1.2.7
On Fri, Mar 04, 2011 at 11:00:00PM +0100, Landry Breuil wrote: :On Fri, Mar 04, 2011 at 03:48:51PM -0600, Marco Peereboom wrote: : It is definitively a regression. A bad one too. Maybe an interaction : with the lib that renders the gifs? : :The only thing i see is https://bugs.webkit.org/show_bug.cgi?id=46437, :which touchs the Gif backend. See if reverting :http://trac.webkit.org/changeset/68446/trunk/WebCore/platform/image-decoders/gif/GIFImageDecoder.cpp :helps.. but i don't think it's the way to go, as it's supposed to fix a :crasher. Backing out http://gitorious.org/webkitgtk/stable/commit/d842bb085aceec4fcfc392a7b76c92c258cd5151 fixes the issue. https://bugs.webkit.org/show_bug.cgi?id=55830 : : On Fri, Mar 04, 2011 at 10:28:55PM +0100, Landry Breuil wrote: : On Fri, Mar 04, 2011 at 07:16:47PM +0100, Dawe wrote: : On Mar 04, 2011 11:22, Marco Peereboom wrote: :It seems that an old webkit bug is back in 1.2.7 where it chokes on :animated gifs. It eats a lot of cpu and sits there blinking instead of :animating. : :For example: http://www.openbsd.org/art/banners/banner1.gif : :Am I the only one seeing that? : : No, I'm seeing it, too. : Tested on amd64. : : A quick googling shows : https://bbs.archlinux.org/viewtopic.php?pid=896669 and : https://bugs.launchpad.net/midori/+bug/721313, but nothing in the : upstream bugtracker. : If it's a regression from 1.2.6 you should check the commits made since : it on http://gitorious.org/webkitgtk, cf the changelog at : http://gitorious.org/webkitgtk/stable/blobs/master/WebKit/gtk/NEWS : : I don't see anything obvious.. : : Landry : : : -- What this world needs is a good five-dollar plasma weapon.
webkit 1.2.7
It seems that an old webkit bug is back in 1.2.7 where it chokes on animated gifs. It eats a lot of cpu and sits there blinking instead of animating. For example: http://www.openbsd.org/art/banners/banner1.gif Am I the only one seeing that?
Re: webkit 1.2.7
On Mar 04, 2011 11:22, Marco Peereboom wrote: It seems that an old webkit bug is back in 1.2.7 where it chokes on animated gifs. It eats a lot of cpu and sits there blinking instead of animating. For example: http://www.openbsd.org/art/banners/banner1.gif Am I the only one seeing that? No, I'm seeing it, too. Tested on amd64.
Re: webkit 1.2.7
On Fri, Mar 04, 2011 at 07:16:47PM +0100, Dawe wrote: On Mar 04, 2011 11:22, Marco Peereboom wrote: It seems that an old webkit bug is back in 1.2.7 where it chokes on animated gifs. It eats a lot of cpu and sits there blinking instead of animating. For example: http://www.openbsd.org/art/banners/banner1.gif Am I the only one seeing that? No, I'm seeing it, too. Tested on amd64. A quick googling shows https://bbs.archlinux.org/viewtopic.php?pid=896669 and https://bugs.launchpad.net/midori/+bug/721313, but nothing in the upstream bugtracker. If it's a regression from 1.2.6 you should check the commits made since it on http://gitorious.org/webkitgtk, cf the changelog at http://gitorious.org/webkitgtk/stable/blobs/master/WebKit/gtk/NEWS I don't see anything obvious.. Landry
Re: webkit 1.2.7
It is definitively a regression. A bad one too. Maybe an interaction with the lib that renders the gifs? On Fri, Mar 04, 2011 at 10:28:55PM +0100, Landry Breuil wrote: On Fri, Mar 04, 2011 at 07:16:47PM +0100, Dawe wrote: On Mar 04, 2011 11:22, Marco Peereboom wrote: It seems that an old webkit bug is back in 1.2.7 where it chokes on animated gifs. It eats a lot of cpu and sits there blinking instead of animating. For example: http://www.openbsd.org/art/banners/banner1.gif Am I the only one seeing that? No, I'm seeing it, too. Tested on amd64. A quick googling shows https://bbs.archlinux.org/viewtopic.php?pid=896669 and https://bugs.launchpad.net/midori/+bug/721313, but nothing in the upstream bugtracker. If it's a regression from 1.2.6 you should check the commits made since it on http://gitorious.org/webkitgtk, cf the changelog at http://gitorious.org/webkitgtk/stable/blobs/master/WebKit/gtk/NEWS I don't see anything obvious.. Landry
Re: webkit 1.2.7
On Fri, Mar 04, 2011 at 03:48:51PM -0600, Marco Peereboom wrote: It is definitively a regression. A bad one too. Maybe an interaction with the lib that renders the gifs? The only thing i see is https://bugs.webkit.org/show_bug.cgi?id=46437, which touchs the Gif backend. See if reverting http://trac.webkit.org/changeset/68446/trunk/WebCore/platform/image-decoders/gif/GIFImageDecoder.cpp helps.. but i don't think it's the way to go, as it's supposed to fix a crasher. On Fri, Mar 04, 2011 at 10:28:55PM +0100, Landry Breuil wrote: On Fri, Mar 04, 2011 at 07:16:47PM +0100, Dawe wrote: On Mar 04, 2011 11:22, Marco Peereboom wrote: It seems that an old webkit bug is back in 1.2.7 where it chokes on animated gifs. It eats a lot of cpu and sits there blinking instead of animating. For example: http://www.openbsd.org/art/banners/banner1.gif Am I the only one seeing that? No, I'm seeing it, too. Tested on amd64. A quick googling shows https://bbs.archlinux.org/viewtopic.php?pid=896669 and https://bugs.launchpad.net/midori/+bug/721313, but nothing in the upstream bugtracker. If it's a regression from 1.2.6 you should check the commits made since it on http://gitorious.org/webkitgtk, cf the changelog at http://gitorious.org/webkitgtk/stable/blobs/master/WebKit/gtk/NEWS I don't see anything obvious.. Landry