User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:9.0.1) Gecko/20100101 Firefox/9.0.1 Build ID: 20111220165912 Steps to reproduce: In our testing we are pulling jpeg images from our camera through our webserver running in Linux (lighttpd) - note the camera is running linux. I am using firefox on my computer connected to the lan where the camera resides. My web application pulls the current jpeg image (1280 X 1024 monochrome) as fast as we can refresh it which is about 3 to 5 fps. We test on released builds so we did not notice this until our firefox went to 8.0. Actual results: The image transitions flicker - they are displayed in a window with a black background so it's not usable with this flicker. Prior to firefox 8.0 the images smoothly transitioned when redraw of next image occurred. Using mozilla regression testing of nightly builds this was introduced on 7/22/11 - because 7/21/11 still produces nice smooth transitions but 7/22/11 introduced what seems like an erase and redraw type of event and the flashing black makes firefox unusable for our product. Expected results: Image transitions should be smooth and you shouldn't see them erased from the web page prior to a redraw of the new image.

Can you please post the "Built from" URLs from about:buildconfig from the last working and the first broken build ? A testcase would be also nice to have...

Matthais I am assuming the "Built from" information would be information that mozregression can report? As for a testcase this is difficult to provide because our camera is not hooked up for connection to outside www. I can try to create a simple case with some images you can place on a webserver and a simple web application but I am in the middle of our initial release so my time is not free but I will do that as soon as I can.

the mozregression tool provides dates and also a URL with the changes in that timeframe. We need this URL or just the Build revisions from both builds.

bug 717572 contains a exact regression range (down to the changeset). marking this one as dupe

I don't think this is a duplicate of bug 717572: the dates do not match. This might be a duplicate of bug 705826, though. The changes in the timeframe given by jcurtis are there: http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2011-07-21&enddate=2011-07-22

You are right, my mistake :-(

I've put together a test case that illustrates the flickering issue. You must put the test case unzipped onto your webserver - then connect to open that over network with mozilla firefox 8.0 or greater - actually this was introduced in the nightly build of 7/22/2011 - anything prior to that date has a smooth transition between images.

Matthais I've attached a test case to this bug report since I was told to create a different report - you've marked duplicate but the test case I provided here will reproduce this issue very clearly.

Thank you very much for the testcase ! I can see the issue at http://matti.no-ip.org/bug717323/ferntest.html I'm pretty sure that it's a dupe, the timeframe matches and you are also replacing the img src=

You're welcome we hope you guys can fix this because we like firefox and right now we can only use safari/chrome and ie. Yes Matthais I agree it is a dupe - Boris told me to put a new ticket out there so I did. In our actual production code we wait for load event callback but it makes no difference with the flicker - so I just put the test case polling 10 times a second. Also the test case I gave you the little play/stop buttons don't stop the timer - sorry bout that just put this code at bottom ferntest.js at end of idle loop if you want to start/stop it. // reset the idle timer if( gAppMgr.playingFern ) { gAppMgr.idleTimerID = setAppTimer(gAppMgr.idleTimerInterval); gAppMgr.idleTimerSet = true; }

I can't reproduce the test using your link Matthais however I did test the same test case locally on the filesystem and the flickering still occurs so you don't have to have a webserver involved.