Bug 35271

Summary: http/tests/media/video-seekable-stall.html fails on the Leopard Commit Queue Machine
Product: WebKit Reporter: Eric Seidel (no email) <eric>
Component: Tools / TestsAssignee: Steve Lacey <sjl>
Status: RESOLVED FIXED    
Severity: Normal CC: abarth, commit-queue, eric.carlson, scherkus, sjl
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: OS X 10.5   
Attachments:
Description Flags
Patch none

Description Eric Seidel (no email) 2010-02-22 16:29:50 PST 
http/tests/media/video-play-stall-seek.html
http/tests/media/video-play-stall.html
http/tests/media/video-seekable-stall.html

--- /tmp/layout-test-results/http/tests/media/video-play-stall-seek-expected.txt	2010-02-22 16:19:37.000000000 -0800
+++ /tmp/layout-test-results/http/tests/media/video-play-stall-seek-actual.txt	2010-02-22 16:19:37.000000000 -0800
@@ -1,11 +1,5 @@
+FAIL: Timed out waiting for notifyDone to be called
 Test that playback can be resumed by seeking backwards after load stalls.
 
 RUN(video.play())
-EVENT(waiting)
-EXPECTED (video.readyState == '2') OK
-RUN(video.currentTime = 0.1)
-EVENT(canplay)
-EXPECTED (video.readyState >= '2') OK
-EXPECTED (video.networkState == '2') OK
-END OF TEST
 
--- /tmp/layout-test-results/http/tests/media/video-play-stall-expected.txt	2010-02-22 16:19:53.000000000 -0800
+++ /tmp/layout-test-results/http/tests/media/video-play-stall-actual.txt	2010-02-22 16:19:53.000000000 -0800
@@ -1,3 +1,4 @@
+FAIL: Timed out waiting for notifyDone to be called
 Test that stalled, timeupdate and waiting events are sent when media load stalls in the middle.
 
 RUN(video.play())
@@ -5,8 +6,6 @@
 EVENT(loadedmetadata)
 EVENT(loadeddata)
 EVENT(canplay)
+EVENT(canplaythrough)
 EVENT(timeupdate)
-EVENT(waiting)
-EVENT(stalled)
-END OF TEST
 
--- /tmp/layout-test-results/http/tests/media/video-seekable-stall-expected.txt	2010-02-22 16:19:53.000000000 -0800
+++ /tmp/layout-test-results/http/tests/media/video-seekable-stall-actual.txt	2010-02-22 16:19:53.000000000 -0800
@@ -5,7 +5,6 @@
 TEST(video.seekable.length == 1) OK
 TEST(video.seekable.start(0) == 0) OK
 TEST(video.seekable.end(0) > 0) OK
-TEST(video.seekable.end(0) < video.duration - 1) OK
-TEST(video.currentTime = video.duration - 1) THROWS(DOMException.INDEX_SIZE_ERR) OK
+TEST(video.seekable.end(0) < video.duration - 1) FAIL
 END OF TEST
 
I'm happy to provide more information about the machine.  Perhaps I'm the only one seeing this.  I'm not sure when this started.
Comment 1 Eric Seidel (no email) 2010-05-12 10:24:31 PDT 
I'm able to reliably reproduce the http/tests/media/video-seekable-stall.html failure.
Comment 2 Eric Seidel (no email) 2010-08-03 15:10:27 PDT 
I've moved the CQ back to that machine again, so now this is a blocking issue for me.  I'll look into skipping the test for just the cq.
Comment 3 Eric Seidel (no email) 2010-08-03 15:11:15 PDT 
I may need to file a new bug.  The one remaining failure is this:

--- /tmp/layout-test-results/http/tests/media/video-seekable-stall-expected.txt	2010-08-03 15:06:49.000000000 -0700
+++ /tmp/layout-test-results/http/tests/media/video-seekable-stall-actual.txt	2010-08-03 15:06:49.000000000 -0700
@@ -5,6 +5,6 @@
 TEST(video.seekable.length == 1) OK
 TEST(video.seekable.start(0) == 0) OK
 TEST(video.seekable.end(0) > 0) OK
-TEST(video.seekable.end(0) < video.duration - 1) OK
+TEST(video.seekable.end(0) < video.duration - 1) FAIL
 END OF TEST
Comment 4 Eric Seidel (no email) 2010-08-03 18:29:52 PDT 
Committed r64611: <http://trac.webkit.org/changeset/64611>
Comment 5 Eric Seidel (no email) 2010-08-03 18:36:22 PDT 
(Slightly regrettably) I've skipped the test for now on Leopard as the most expedient solution.  I'm certain my machine is not the only one seeing this failure.
Comment 6 Steve Lacey 2011-05-31 15:44:03 PDT 
I'm working on a related bug with video-seekable-stall for chromium and have an issue with the last check in the test:

TEST(video.seekable.end(0) < video.duration - 1) OK

Why the "-1"? The max time range, "video.seekable.end(0)" is defined to be video.duration plus the time of the first frame in the video (which is zero)... 

http://www.w3.org/TR/html5/video.html#dom-media-seekable

With the "-1" removed, this will work in chromium... (when I've checked in another fix)...
Comment 7 Eric Carlson 2011-05-31 15:53:12 PDT 
(In reply to comment #6)
> I'm working on a related bug with video-seekable-stall for chromium and have an issue with the last check in the test:
> 
> TEST(video.seekable.end(0) < video.duration - 1) OK
> 
> Why the "-1"? The max time range, "video.seekable.end(0)" is defined to be video.duration plus the time of the first frame in the video (which is zero)... 
> 
> http://www.w3.org/TR/html5/video.html#dom-media-seekable
> 
> With the "-1" removed, this will work in chromium... (when I've checked in another fix)...

The test was written at a time when the only media engine, QuickTime, could only seek to a time that was already loaded so checking that the last range was before "duration - 1" was correct because of the stallAt=100000.

This behavior is still true of QuickTime on Windows, so removing the "-1" will make it fail.

I think this test has outlived its usefulness
Comment 8 Steve Lacey 2011-05-31 16:02:55 PDT 
It's currently skipped on windows. I'll send a patch to remove the test.
Comment 9 Steve Lacey 2011-05-31 16:15:23 PDT 
Created attachment 95503 [details]
Patch
Comment 10 Eric Seidel (no email) 2011-05-31 18:50:34 PDT 
I"m OK with this if Eric is.
Comment 11 WebKit Commit Bot 2011-06-01 09:25:34 PDT 
The commit-queue encountered the following flaky tests while processing attachment 95503 [details]:

http/tests/websocket/tests/error-detect.html bug 54012 (author: abarth@webkit.org)
The commit-queue is continuing to process your patch.
Comment 12 WebKit Commit Bot 2011-06-01 09:27:34 PDT 
Comment on attachment 95503 [details]
Patch

Clearing flags on attachment: 95503

Committed r87817: <http://trac.webkit.org/changeset/87817>
Comment 13 WebKit Commit Bot 2011-06-01 09:27:39 PDT 
All reviewed patches have been landed.  Closing bug.