Layout test for window.devicePixelRatio
Created attachment 130141 [details] Patch
Ossy, we are worried about that the level of precision the test introduces could have different results on ARM. Mind to help us to verify that on QtWebKit arm bots? We do not want to red-ify it. Thanks
(In reply to comment #2) > Ossy, we are worried about that the level of precision the test introduces could have different results on ARM. > > Mind to help us to verify that on QtWebKit arm bots? We do not want to red-ify it. Thanks Don't worry! ;) I suggested to Konrad on IRC that you should use a given +/- epsilon tolerance instead of expecting same high precision numbers on all platforms. Or you can land it and check what bots say and then adjust the epsilion to make all platform happy.
Created attachment 130846 [details] Patch
(In reply to comment #3) > (In reply to comment #2) > > Ossy, we are worried about that the level of precision the test introduces could have different results on ARM. > > > > Mind to help us to verify that on QtWebKit arm bots? We do not want to red-ify it. Thanks > > Don't worry! ;) I suggested to Konrad on IRC that you should > use a given +/- epsilon tolerance instead of expecting same > high precision numbers on all platforms. The latest patch does just that by using shouldBeCloseTo > > Or you can land it and check what bots say and then adjust the > epsilion to make all platform happy.
Comment on attachment 130846 [details] Patch LGTM. Let's see the results on other platforms :)
Comment on attachment 130846 [details] Patch Clearing flags on attachment: 130846 Committed r110197: <http://trac.webkit.org/changeset/110197>
All reviewed patches have been landed. Closing bug.
This revision (r110197) causes fail on Qt and on GTK. The diffs are the same: --- /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/viewport/viewport-133-expected.txt +++ /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/viewport/viewport-133-actual.txt @@ -1,3 +1,3 @@ viewport size 480x534 scale 1.000000 with limits [1.000000, 5.000000] and userScalable -1.000000 -PASS window.devicePixelRatio is within 0.001 of 0.7083333134651184 +FAIL window.devicePixelRatio should be within 0.001 of 0.7083333134651184. Was 1.
Reopen it.
*** Bug 80689 has been marked as a duplicate of this bug. ***
(In reply to comment #9) > This revision (r110197) causes fail on Qt and on GTK. The diffs are the same: > > --- /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/viewport/viewport-133-expected.txt > +++ /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/viewport/viewport-133-actual.txt > @@ -1,3 +1,3 @@ > viewport size 480x534 scale 1.000000 with limits [1.000000, 5.000000] and userScalable -1.000000 > -PASS window.devicePixelRatio is within 0.001 of 0.7083333134651184 > +FAIL window.devicePixelRatio should be within 0.001 of 0.7083333134651184. Was 1. This test is related to https://bugs.webkit.org/show_bug.cgi?id=80160 where I set the devicePixelRatio based based on the results from computeViewportAttributes. This should work on Qt since I believe that they also set this value in the same way. So the failure could be a genuine failure for Qt. Looking further into other ports I see now the EFL and GTK don't do this so I'll post a patch soon that will add this new test to the skipped list for those ports.
Created attachment 131029 [details] Patch
Comment on attachment 130846 [details] Patch webkit-patch incorrectly marked this as obsolete
Comment on attachment 131029 [details] Patch Clearing flags on attachment: 131029 Committed r110294: <http://trac.webkit.org/changeset/110294>
We're considering changing the behavior tested by this test in https://bugs.webkit.org/show_bug.cgi?id=87407