Closed Bug 975270 Opened 6 years ago Closed 6 years ago
abort and emptied events should not be sync during load()
User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/33.0.1750.58 Safari/537.36 OPR/20.0.1387.30 (Edition Next) Steps to reproduce: I was reading HTMLMediaElement.cpp and thought mIsRunningLoadMethod looked suspicious, so I wrote this test case for you: https://github.com/w3c/web-platform-tests/pull/661 Actual results: The abort and emptied events are fired sync: DispatchEvent(NS_LITERAL_STRING("abort")); Expected results: Should be fired in a queued task: DispatchAsyncEvent(NS_LITERAL_STRING("abort"));
Good spotting. Let's see if this flushes out any test failures: https://tbpl.mozilla.org/?tree=Try&rev=0cd1fd70915a
Trying again now that I should have fixed the test failres that showed up: https://tbpl.mozilla.org/?tree=Try&rev=5aae635d6c3e
Assignee: nobody → cpearce
Status: UNCONFIRMED → ASSIGNED
Ever confirmed: true
Attachment #8384095 - Flags: review?(roc)
Comment on attachment 8384095 [details] [diff] [review] Patch: Dispatch "emptied" and "abort" async, as they should be per the spec Review of attachment 8384095 [details] [diff] [review]: ----------------------------------------------------------------- Test?
Attachment #8384095 - Flags: review?(roc) → review+
(I added a test and included that when landing.)
had to backout this changes for a permanent failure in b2g-ics opt builds like https://tbpl.mozilla.org/php/getParsedLog.php?id=35520278&tree=Mozilla-Inbound
Ooops, I forgot that on b2g we default prelaoad to "none", so the test would never reach loadedmetadata. I'll fix the test.
Once more, fixed test: https://hg.mozilla.org/integration/mozilla-inbound/rev/e5ae3b72e03c
Status: ASSIGNED → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla30
You need to log in before you can comment on or make changes to this bug.