For a first pass, we would have to assume that all painting from MSE taints the canvas.
Second pass, taint the canvas if any non-single-origin XHR requests were made.
Third pass, attach the single-origin state to the ArrayBuffer generated by XHRs, and ferry that data through MSE to the video element.
This is a matter of leaving things to the last minute. An accident waiting to happen.
It is related to the CORS flaw. Once CORS support is enabled , MediaSource WebGL textures will still be a problem. A very basic usage.
See this for reference
https://bugs.webkit.org/show_bug.cgi?id=135379
My proxy hack work around does not work on Dash only mp4 files.
I got myself confused. It seems when drawing to 2D canvas there is no CORS restriction it obviously does not use CORS. It's only when you attempt to get a data uri to get an image from the video does CORS become a problem which requires the proxy hack.
However when trying to use WebGL there is and because Safari lacks CORS support , we have this issue.
Here is a raw webgl example just with an mp4 file.
https://jsfiddle.net/7t77rz6L/11/
"SecurityError: DOM Exception 18: An attempt was made to break through the security policy of the user agent."
Just confirming even with the proxy hack, Iphone will not play inline. It needs serious dodgy hacks of updating the loaded video time and separate audio. No good for webgl / VR needing to display the canvas.
That same proxy hack to work around CORS does not work on MediaSource however.
Attachment 272504[details] did not pass style-queue:
ERROR: Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770: Place brace on its own line for function definitions. [whitespace/braces] [4]
Total errors found: 1 in 12 files
If any of these errors are false positives, please file a bug against check-webkit-style.
That dependancy bug is fixed. Does that means CORS and webgl will play nicely soon ?
Absolutely everyone in VR is depending on this minus the Iphone inline issue, not sure how to tackle that one, some nasty hacks out there.
Apple can't market VR until this is done really but Iphone still unusable.
(In reply to comment #12)
> That dependancy bug is fixed. Does that means CORS and webgl will play
> nicely soon ?
This bug will only address the painting to canvas use case. There is still more work to enable efficient rendering to WebGL.
Attachment 272565[details] did not pass style-queue:
ERROR: Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770: Place brace on its own line for function definitions. [whitespace/braces] [4]
Total errors found: 1 in 12 files
If any of these errors are false positives, please file a bug against check-webkit-style.
Attachment 272566[details] did not pass style-queue:
ERROR: Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770: Place brace on its own line for function definitions. [whitespace/braces] [4]
Total errors found: 1 in 12 files
If any of these errors are false positives, please file a bug against check-webkit-style.
Created attachment 272568[details]
Archive of layout-test-results from ews103 for mac-yosemite
The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews103 Port: mac-yosemite Platform: Mac OS X 10.10.5
Created attachment 272570[details]
Archive of layout-test-results from ews104 for mac-yosemite-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews104 Port: mac-yosemite-wk2 Platform: Mac OS X 10.10.5
Crashes are an ASSERT in initVTDecompressionSessionDecodeFrameWithOutputHandler(...). Looking at the headers, I see __OSX_AVAILABLE_STARTING(__MAC_10_11,__IPHONE_9_0);
Looks like I'll need to move things around to use VTDecompressionSessionDecodeFrame instead.
Created attachment 272572[details]
Archive of layout-test-results from ews112 for mac-yosemite
The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews112 Port: mac-yosemite Platform: Mac OS X 10.10.5
Comment on attachment 272657[details]
Patch
View in context: https://bugs.webkit.org/attachment.cgi?id=272657&action=review
r=me wth some nits and restructuring suggestions.
> Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:580
> + m_player->firstVideoFrameAvailable();
Do you want to call firstVideoFrameAvailable more than once in the player's lifecycle?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:611
> +class WebCoreDecompressionSession : public ThreadSafeRefCounted<WebCoreDecompressionSession> {
> +public:
> + static Ref<WebCoreDecompressionSession> create(AVSampleBufferDisplayLayer* layer, AVSampleBufferRenderSynchronizer *synchronizer) { return adoptRef(*new WebCoreDecompressionSession(layer, synchronizer)); }
Is there any reason to hide this here in SourceBufferPrivateAVFObjC.mm? It doesn't look like this class is MSE-specific in any way, and it could be extremely useful for other media engines (eg. MediaStream).
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:650
> + std::function<void()> m_notificationCallback;
Nit: This name is really generic for a callback with a specific purpose. Maybe m_requestMediaDataCallback?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:700
> + timingInfoArray.grow(itemCount);
Why .grow() instead of .reserveCapacity(), or Vector<CMSampleTimingInfo> vector(itemCount)?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:720
> + CMBufferQueueCreate(kCFAllocatorDefault, kMaximumCapacity, &callbacks, &outQueue);
Nit: kMaximumQueueCapacity would be clearer.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:723
> + CMBufferQueueInstallTriggerWithIntegerThreshold(m_producerQueue.get(), maybeBecomeReadyForMoreMediaDataCallback, this, kCMBufferQueueTrigger_WhenBufferCountBecomesLessThan, 15, &m_didBecomeReadyTrigger);
This magic number should be a named constant to make it easier to understand what it means.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:760
> + if (m_decompressionSession && !VTDecompressionSessionCanAcceptFormatDescription(m_decompressionSession.get(), videoFormatDescription)) {
> + VTDecompressionSessionWaitForAsynchronousFrames(m_decompressionSession.get());
> + m_decompressionSession = nullptr;
> + }
Nit: you should definitely add error logging here to make it easier to diagnose failures.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770
> + VTDecompressionOutputCallbackRecord callback {
> + &decompressionOutputCallback,
> + this,
> + };
What guarantees that "this" stays alive? Should it be a weakPtr?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:794
> + if (noErr != CMVideoFormatDescriptionCreateForImageBuffer(kCFAllocatorDefault, rawImageBuffer, &rawImageBufferDescription))
> + return;
You should log this error.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:805
> + if (noErr != CMSampleBufferCreateForImageBuffer(kCFAllocatorDefault, rawImageBuffer, true, nullptr, nullptr, imageBufferDescription.get(), &imageBufferTiming, &rawImageSampleBuffer))
> + return;
Ditto.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:809
> + callOnMainThread([strongThis, status, imageSampleBuffer, infoFlags, displaying] {
> + UNUSED_PARAM(infoFlags);
Why pass infoFlags if you don't use them?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:819
> + --m_framesBeingDecoded;
ASSERT(m_framesBeingDecoded >= 0)
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:858
> + RetainPtr<CMSampleBufferRef> sample = adoptCF((CMSampleBufferRef)CMBufferQueueDequeueAndRetain(m_producerQueue.get()));
> + [m_displayLayer enqueueSampleBuffer:sample.get()];
> + CMBufferQueueEnqueue(m_consumerQueue.get(), sample.get());
Is this correct if [m_displayLayer isReadyForMoreMediaData] above returned false?
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:935
> + CMSampleBufferRef sample = (CMSampleBufferRef)(buf);
> + return CMSampleBufferGetDecodeTimeStamp(sample);
Nit: you don't need the local variable.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:942
> + CMSampleBufferRef sample = (CMSampleBufferRef)(buf);
> + return CMSampleBufferGetPresentationTimeStamp(sample);
Ditto.
> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:949
> + CMSampleBufferRef sample = (CMSampleBufferRef)(buf);
> + return CMSampleBufferGetDuration(sample);
Ditto.
Comment on attachment 272657[details]
Patch
View in context: https://bugs.webkit.org/attachment.cgi?id=272657&action=review>> Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:580
>> + m_player->firstVideoFrameAvailable();
>
> Do you want to call firstVideoFrameAvailable more than once in the player's lifecycle?
Yes. This lets the media element know to switch from displaying the poster frame to displaying the video. And this is not limited to happening only once per player. If this player is seeked out of it's buffered range, it could not have flushed all its images and have nothing to display, in which case the media element will have switched back to displaying the poster. Perhaps we need to rename the client method though.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:611
>> + static Ref<WebCoreDecompressionSession> create(AVSampleBufferDisplayLayer* layer, AVSampleBufferRenderSynchronizer *synchronizer) { return adoptRef(*new WebCoreDecompressionSession(layer, synchronizer)); }
>
> Is there any reason to hide this here in SourceBufferPrivateAVFObjC.mm? It doesn't look like this class is MSE-specific in any way, and it could be extremely useful for other media engines (eg. MediaStream).
This could get pulled into its own file, sure.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:650
>> + std::function<void()> m_notificationCallback;
>
> Nit: This name is really generic for a callback with a specific purpose. Maybe m_requestMediaDataCallback?
Sure.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:700
>> + timingInfoArray.grow(itemCount);
>
> Why .grow() instead of .reserveCapacity(), or Vector<CMSampleTimingInfo> vector(itemCount)?
The latter isn't possible; you need to pass in a `const CMSampleTimingInfo&` in addition to a itemCount, and setting specific values is unnecessary since the next method will overwrite the contents anyway. And while the former resizes the Vector's buffer, it doesn't resize the vector itself.
Of course, now that I read this, I can't remember why I needed the sample timing at all. I'm pretty sure I can drop this whole section.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:720
>> + CMBufferQueueCreate(kCFAllocatorDefault, kMaximumCapacity, &callbacks, &outQueue);
>
> Nit: kMaximumQueueCapacity would be clearer.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:723
>> + CMBufferQueueInstallTriggerWithIntegerThreshold(m_producerQueue.get(), maybeBecomeReadyForMoreMediaDataCallback, this, kCMBufferQueueTrigger_WhenBufferCountBecomesLessThan, 15, &m_didBecomeReadyTrigger);
>
> This magic number should be a named constant to make it easier to understand what it means.
You're right, this is supposed to be kLowWaterMark.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:760
>> + }
>
> Nit: you should definitely add error logging here to make it easier to diagnose failures.
This is definitely not an error condition. If the client starts pushing in different dimension samples, we may need to restart the decompression session.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770
>> + };
>
> What guarantees that "this" stays alive? Should it be a weakPtr?
Unfortunately, it can't be. It has to be a bare pointer. However, we're not using asynchronous decoding flags when we decode frames, so the callback will occur before VTDecompressionSessionDecodeFrame() returns. So there's no danger here.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:794
>> + return;
>
> You should log this error.
ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:805
>> + return;
>
> Ditto.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:809
>> + UNUSED_PARAM(infoFlags);
>
> Why pass infoFlags if you don't use them?
Good point. :)
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:819
>> + --m_framesBeingDecoded;
>
> ASSERT(m_framesBeingDecoded >= 0)
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:858
>> + CMBufferQueueEnqueue(m_consumerQueue.get(), sample.get());
>
> Is this correct if [m_displayLayer isReadyForMoreMediaData] above returned false?
This won't be reached if -isReadyForMoreMediaData returns false.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:935
>> + return CMSampleBufferGetDecodeTimeStamp(sample);
>
> Nit: you don't need the local variable.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:942
>> + return CMSampleBufferGetPresentationTimeStamp(sample);
>
> Ditto.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:949
>> + return CMSampleBufferGetDuration(sample);
>
> Ditto.
Ok.
Comment on attachment 272657[details]
Patch
View in context: https://bugs.webkit.org/attachment.cgi?id=272657&action=review>> Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:580
>> + m_player->firstVideoFrameAvailable();
>
> Do you want to call firstVideoFrameAvailable more than once in the player's lifecycle?
Yes. This lets the media element know to switch from displaying the poster frame to displaying the video. And this is not limited to happening only once per player. If this player is seeked out of it's buffered range, it could not have flushed all its images and have nothing to display, in which case the media element will have switched back to displaying the poster. Perhaps we need to rename the client method though.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:611
>> + static Ref<WebCoreDecompressionSession> create(AVSampleBufferDisplayLayer* layer, AVSampleBufferRenderSynchronizer *synchronizer) { return adoptRef(*new WebCoreDecompressionSession(layer, synchronizer)); }
>
> Is there any reason to hide this here in SourceBufferPrivateAVFObjC.mm? It doesn't look like this class is MSE-specific in any way, and it could be extremely useful for other media engines (eg. MediaStream).
This could get pulled into its own file, sure.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:650
>> + std::function<void()> m_notificationCallback;
>
> Nit: This name is really generic for a callback with a specific purpose. Maybe m_requestMediaDataCallback?
Sure.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:700
>> + timingInfoArray.grow(itemCount);
>
> Why .grow() instead of .reserveCapacity(), or Vector<CMSampleTimingInfo> vector(itemCount)?
The latter isn't possible; you need to pass in a `const CMSampleTimingInfo&` in addition to a itemCount, and setting specific values is unnecessary since the next method will overwrite the contents anyway. And while the former resizes the Vector's buffer, it doesn't resize the vector itself.
Of course, now that I read this, I can't remember why I needed the sample timing at all. I'm pretty sure I can drop this whole section.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:720
>> + CMBufferQueueCreate(kCFAllocatorDefault, kMaximumCapacity, &callbacks, &outQueue);
>
> Nit: kMaximumQueueCapacity would be clearer.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:723
>> + CMBufferQueueInstallTriggerWithIntegerThreshold(m_producerQueue.get(), maybeBecomeReadyForMoreMediaDataCallback, this, kCMBufferQueueTrigger_WhenBufferCountBecomesLessThan, 15, &m_didBecomeReadyTrigger);
>
> This magic number should be a named constant to make it easier to understand what it means.
You're right, this is supposed to be kLowWaterMark.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:760
>> + }
>
> Nit: you should definitely add error logging here to make it easier to diagnose failures.
This is definitely not an error condition. If the client starts pushing in different dimension samples, we may need to restart the decompression session.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:770
>> + };
>
> What guarantees that "this" stays alive? Should it be a weakPtr?
Unfortunately, it can't be. It has to be a bare pointer. However, we're not using asynchronous decoding flags when we decode frames, so the callback will occur before VTDecompressionSessionDecodeFrame() returns. So there's no danger here.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:794
>> + return;
>
> You should log this error.
ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:805
>> + return;
>
> Ditto.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:809
>> + UNUSED_PARAM(infoFlags);
>
> Why pass infoFlags if you don't use them?
Good point. :)
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:819
>> + --m_framesBeingDecoded;
>
> ASSERT(m_framesBeingDecoded >= 0)
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:858
>> + CMBufferQueueEnqueue(m_consumerQueue.get(), sample.get());
>
> Is this correct if [m_displayLayer isReadyForMoreMediaData] above returned false?
This won't be reached if -isReadyForMoreMediaData returns false.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:935
>> + return CMSampleBufferGetDecodeTimeStamp(sample);
>
> Nit: you don't need the local variable.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:942
>> + return CMSampleBufferGetPresentationTimeStamp(sample);
>
> Ditto.
Ok.
>> Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:949
>> + return CMSampleBufferGetDuration(sample);
>
> Ditto.
Ok.
Hi guys this has gone quiet. VR people are waiting for this fix and it's pretty critical. Already missed the boat it seems !
I'm going to see if running in WebView / Cordova works around this lack of implementation like it can override inline restrictions.
Any ideas of the status of this ?
I know it's pretty much crept up and steamrolled, many people trying to do WebVR including Youtube and Facebook of all people are also waiting.
They should pay to make this happen.
This problem has been left exposed for many years now and was just an accident waiting to happen. For some reason a few of us have to do the run around for many others.
Obviously the reverse proxy work around is not ok for production so not even Facebook or Youtube are doing that and therefore Safari is broken.
If there is a nightly fix to test please let me know.
Hi guys I'm sorry to use this as a forum. But by stealth it seems Apple have forked the code and fixed it correct ?
This is not so useful for current browsers so wondering if they will merge it back ?
https://twitter.com/zenoc/status/742770789880111104
(In reply to comment #36)
> Hi guys I'm sorry to use this as a forum. But by stealth it seems Apple have
> forked the code and fixed it correct ?
>
> This is not so useful for current browsers so wondering if they will merge
> it back ?
>
> https://twitter.com/zenoc/status/742770789880111104
This only applies to videos on macOS running Sierra.
Yes I am aware they are only interested in fixing it in an OS nobody even has yet. Which renders all other OS useless.
However I've discovered a quiet change in Safari in Yosemite which has broken feature detection.
Something truly bizarre has changed in Safari on Yosemite that was not there before. It has broken feature detection with CORS in alot of things I have done.
Just a heads up people just working how broken it really is.
```
var testVideo = document.createElement("video");
testVideo.crossOrigin = "anonymous";
console.log(testVideo.hasAttribute("crossOrigin"));
```
In IE11 it returns false as expected. In Safari it's true but the CORS bug is still obviously there. The feature detection is there to obviously require the proxy hack. Madness.
So I can't help to think some half baked change was pushed ?
Created attachment 306704[details]
Archive of layout-test-results from ews100 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 306706[details]
Archive of layout-test-results from ews123 for ios-simulator-wk2
The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews123 Port: ios-simulator-wk2 Platform: Mac OS X 10.11.6
Created attachment 306707[details]
Archive of layout-test-results from ews104 for mac-elcapitan-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews104 Port: mac-elcapitan-wk2 Platform: Mac OS X 10.11.6
Created attachment 306709[details]
Archive of layout-test-results from ews113 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews113 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 306725[details]
Archive of layout-test-results from ews102 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews102 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 306726[details]
Archive of layout-test-results from ews106 for mac-elcapitan-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews106 Port: mac-elcapitan-wk2 Platform: Mac OS X 10.11.6
Created attachment 306730[details]
Archive of layout-test-results from ews113 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews113 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 306744[details]
Archive of layout-test-results from ews100 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 306746[details]
Archive of layout-test-results from ews107 for mac-elcapitan-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews107 Port: mac-elcapitan-wk2 Platform: Mac OS X 10.11.6
Created attachment 306747[details]
Archive of layout-test-results from ews116 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews116 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 307148[details]
Archive of layout-test-results from ews100 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 307151[details]
Archive of layout-test-results from ews105 for mac-elcapitan-wk2
The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews105 Port: mac-elcapitan-wk2 Platform: Mac OS X 10.11.6
Created attachment 307153[details]
Archive of layout-test-results from ews116 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews116 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Comment on attachment 310564[details]
Patch for landing
Rejecting attachment 310564[details] from commit-queue.
Failed to run "['/Volumes/Data/EWS/WebKit/Tools/Scripts/webkit-patch', '--status-host=webkit-queues.webkit.org', '--bot-id=webkit-cq-03', 'build', '--no-clean', '--no-update', '--build-style=release', '--port=mac']" exit_code: 2 cwd: /Volumes/Data/EWS/WebKit
Last 500 characters of output:
d/Objects-normal/x86_64/VideoToolboxSoftLink.o
clang: error: no such file or directory: '/Volumes/Data/EWS/WebKit/Source/WebCore/platform/cocoa/VideoToolboxSoftLink.cpp'
clang: error: no input files
** BUILD FAILED **
The following build commands failed:
CompileC /Volumes/Data/EWS/WebKit/WebKitBuild/WebCore.build/Release/WebCore.build/Objects-normal/x86_64/VideoToolboxSoftLink.o platform/cocoa/VideoToolboxSoftLink.cpp normal x86_64 c++ com.apple.compilers.llvm.clang.1_0.compiler
(1 failure)
Full output: http://webkit-queues.webkit.org/results/3772093
Comment on attachment 310586[details]
Patch for landing
Rejecting attachment 310586[details] from commit-queue.
New failing tests:
imported/w3c/web-platform-tests/media-source/mediasource-config-change-mp4-a-bitrate.html
media/media-source/media-source-remove.html
media/media-source/media-source-fastseek.html
media/media-source/media-source-sequence-timestamps.html
imported/w3c/web-platform-tests/media-source/mediasource-avtracks.html
Full output: http://webkit-queues.webkit.org/results/3772975
Created attachment 310591[details]
Archive of layout-test-results from webkit-cq-03 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the commit-queue.
Bot: webkit-cq-03 Port: mac-elcapitan Platform: Mac OS X 10.11.6
Created attachment 310611[details]
Archive of layout-test-results from webkit-cq-03 for mac-elcapitan
The attached test failures were seen while running run-webkit-tests on the commit-queue.
Bot: webkit-cq-03 Port: mac-elcapitan Platform: Mac OS X 10.11.6
I might be doing something wrong here, but this doesn't seem to work on Safari 13.
Made a simple test page here:
https://s3.amazonaws.com/storage2.interlude.fm/dev_temp/tomer/safari_mse_canvas_bug/index.html
The above test page works correctly on other browsers (including IE), but MSE video is not drawn onto the canvas on Safari 13 (desktop + iPadOS).
Would appreciate any pointers - thanks!!
(In reply to tomerlahav from comment #97)
> I might be doing something wrong here, but this doesn't seem to work on
> Safari 13.
> Made a simple test page here:
> https://s3.amazonaws.com/storage2.interlude.fm/dev_temp/tomer/
> safari_mse_canvas_bug/index.html
>
> The above test page works correctly on other browsers (including IE), but
> MSE video is not drawn onto the canvas on Safari 13 (desktop + iPadOS).
> Would appreciate any pointers - thanks!!
Would you mind filing a new bug? It is hard to track the bug in an issue that was resolved.
Update to this bug: it was rolled out because it caused a large memory regression in our performance tests. So we're back to the status-quo-ante, where painting of MSE content when the element is in the DOM is broken. The current best workaround is to remove the element from the DOM if you want to paint it to a canvas. Apologies for the late response here.
I don't have a use case where MSE is in use as HLS falls back to native HLS. My target for Safari is native HLS. I am finding sometimes HLS refuses to paint at all until another page reload. On preload frame is black then continues like that. On Ipad IOS 15 this time.
Would removing video from the dom help here too ?
I played around with this and doing this with all browsers Chrome is pausing upon removal of the element for certain situations. MSE and mediastream for one. sad.
2016-02-29 11:29 PST, Jer Noble
2016-03-01 08:45 PST, Jer Noble
2016-03-01 09:17 PST, Jer Noble
2016-03-01 10:02 PST, Build Bot
2016-03-01 10:06 PST, Build Bot
2016-03-01 10:20 PST, Build Bot
2016-03-01 11:13 PST, Jer Noble
2016-03-02 08:27 PST, Jer Noble
2016-12-08 10:43 PST, Jer Noble
2016-12-08 10:49 PST, Jer Noble
2017-03-21 16:36 PDT, Jer Noble
2017-04-10 08:55 PDT, Jer Noble
2017-04-10 10:31 PDT, Build Bot
2017-04-10 10:33 PDT, Build Bot
2017-04-10 10:36 PDT, Build Bot
2017-04-10 10:45 PDT, Build Bot
2017-04-10 11:05 PDT, Jer Noble
2017-04-10 11:58 PDT, Build Bot
2017-04-10 12:05 PDT, Build Bot
2017-04-10 12:30 PDT, Build Bot
2017-04-10 13:21 PDT, Jer Noble
2017-04-10 14:39 PDT, Build Bot
2017-04-10 14:44 PDT, Build Bot
2017-04-10 14:48 PDT, Build Bot
2017-04-14 12:37 PDT, Jer Noble
2017-04-14 13:29 PDT, Jer Noble
2017-04-14 14:47 PDT, Build Bot
2017-04-14 15:07 PDT, Build Bot
2017-04-14 15:20 PDT, Build Bot
2017-05-18 15:49 PDT, Jer Noble
2017-05-18 16:28 PDT, Jer Noble
2017-05-18 18:22 PDT, Jer Noble
2017-05-18 19:07 PDT, WebKit Commit Bot
2017-05-18 20:47 PDT, Jer Noble
2017-05-18 21:55 PDT, WebKit Commit Bot
2017-05-18 22:10 PDT, Jer Noble
2017-05-19 22:19 PDT, Jer Noble