RESOLVED FIXED 16902
fast/encoding/mailto-always-utf-8.html fails when run after fast/dom/Window/window-property-shadowing.html
https://bugs.webkit.org/show_bug.cgi?id=16902
Summary fast/encoding/mailto-always-utf-8.html fails when run after fast/dom/Window/w...
mitz
Reported 2008-01-16 23:12:04 PST
run-webkit-tests fast/dom/Window/window-property-shadowing.html fast/encoding/mailto-always-utf-8.html produces the failure seen in <http://build.webkit.org/results/trunk-win-release/7072/fast/encoding/mailto-always-utf-8-diffs.txt>. (I think the failure is not seen on Mac build bots simply because the tests are not in the same 1000-batch on Mac).
Attachments
reduced test case (170 bytes, text/html)
2008-01-17 00:26 PST, Alexey Proskuryakov
no flags
proposed fix (33.52 KB, patch)
2008-01-17 10:48 PST, Alexey Proskuryakov
darin: review+
Alexey Proskuryakov
Comment 1 2008-01-17 00:13:02 PST
This looks like a real bug, not just a DRT issue.
Alexey Proskuryakov
Comment 2 2008-01-17 00:26:22 PST
Created attachment 18493 [details] reduced test case
Alexey Proskuryakov
Comment 3 2008-01-17 01:25:20 PST
Same problem with window.devicePixelRatio, other properties seem to be correct.
Alexey Proskuryakov
Comment 4 2008-01-17 01:36:21 PST
Please ignore the previous comment - devicePixelRatio is OK, too, so it's only the name that isn't reset.
Adam Roben (:aroben)
Comment 5 2008-01-17 09:11:31 PST
Alexey Proskuryakov
Comment 6 2008-01-17 09:12:15 PST
When running an extended test in DRT, I'm also getting an exception when trying to access window.showModalDialog in a test running after fast/dom/Window/window-property-shadowing.html. Somehow, this doesn't happen in Safari.
Adam Roben (:aroben)
Comment 7 2008-01-17 10:44:29 PST
Added to the Windows Skipped file in r29584.
Alexey Proskuryakov
Comment 8 2008-01-17 10:48:18 PST
Created attachment 18504 [details] proposed fix The showModalDialog exception is correct behavior - that's because DRT cannot run modal, and says so via its UIDelegate. There are many places in code where I could put the fix into, not sure if I picked the best (or even correct) one.
Darin Adler
Comment 9 2008-01-17 11:25:35 PST
Comment on attachment 18504 [details] proposed fix I think you could come up with some other way to name the new test differently. Adding an underscore seems pretty strange. r=me
Alexey Proskuryakov
Comment 10 2008-01-17 11:33:38 PST
(In reply to comment #9) > I think you could come up with some other way to name the new test differently. > Adding an underscore seems pretty strange. I have tried "-2" and "2", but both ran before the original test. I don't like the underscore, too, but couldn't find a better option.
Alexey Proskuryakov
Comment 11 2008-01-17 12:07:36 PST
Committed revision 29590.
Note You need to log in before you can comment on or make changes to this bug.