Created attachment 8879321 [details] Auswahl_120.png User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:56.0) Gecko/20100101 Firefox/56.0 Build ID: 20170619100247 Steps to reproduce: Go to https://web.whatsapp.com/ as being logged in Actual results: Around the outer edges of the content I see layout artefacts. I do not think these happened when I still used Dev Edition (I just recently move to Nightly) Expected results: FWIW, as soon as I go to another tab and then back to https://web.whatsapp.com/ – or change the window size; the focus changes – then these artefacts are gone. Happens always again be just pressing F5 to reload.
Gerd, do you see these layout artifacts with the current release 54 too? If not, it could be a regression in 56. In that case, could you use the tool mozregression to narrow down a regression range. See http://mozilla.github.io/mozregression/ for details. Run the command "mozregression --good=54" then for each nightly build launched, make the test and enter if it's good or bad. After the run, paste here the final pushlog.
Loic, here you go: 22:30.15 INFO: Narrowed inbound regression window from [958d2a5d, 0ff83c9d] (3 revisions) to [958d2a5d, 7f894f79] (2 revisions) (~1 steps left) 22:30.15 INFO: Oh noes, no (more) inbound revisions :( 22:30.15 INFO: Last good revision: 958d2a5d10091401fd5e900e8e063d21940c137e 22:30.15 INFO: First bad revision: 7f894f791cdf170d788507d0eff30024ce699523 22:30.15 INFO: Pushlog: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=958d2a5d10091401fd5e900e8e063d21940c137e&tochange=7f894f791cdf170d788507d0eff30024ce699523 22:31.10 INFO: Looks like the following bug has the changes which introduced the regression: https://bugzilla.mozilla.org/show_bug.cgi?id=1370430 Let me know if you need anything more. (FWIW, I the layout breakage is around the outer spaces of https://web.whatsapp.com and this is only visible when the window width is bigger than say 1600px)
There are these patches: Matt Woodrow — Bug 1359709 - Scale visible region for preserve-3d layers correctly. r=thinker Matt Woodrow — Bug 1359709 - Use the DOM-ordering parent frame when deciding if a frame combines its transform with ancestors. r=dbaron Matt Woodrow — Bug 1361970 - Make PostProcessLayers occlusion culling work against the surface we will draw to rather than the parent layer. r=mstange Matt, do you think your patches could have regressed this?
This might be a dup of bug 1373335, do the builds from the try push there fix it?
I guess you mean https://treeherder.mozilla.org/#/jobs?repo=try&revision=bd8be33db1e07d8fcc2ebcb25b027bcc802fa6f0 (as said in comment no 8 of bug 1373335). Hmm, well, I could test it if I knew how ;) Just spent 10min to find a download link under this treeherder page... Searching the web I came up with http://email@example.com/ but this only has win+mac builds and I am on Linux. Loic, any hints or are there docs on running the builds from the try push?
Created attachment 8879905 [details] 2017-06-21 13_07_53-Mozregression-gui.png I guess I was able to run the aforementioned try build with mozregression. See screenshot, I hope I picked the right build... With this build the bug is definitely still visible. Hence, I'm re-NIing to Matt.
As of 56.0a1 (2017-06-23) (64-bit) this seems to be fixed. My guess would be that this might have been a dup of Bug 1373479 which is also fixed as of today.
Let's dupe it.
Fixed in 56 in the duplicate.