Bug 164561 – [MediaStream] apply constraints passed to getUserMedia()

• Component: Media
• Status: RESOLVED FIXED

[Eric Carlson]

Restructure the way constraints passed to getUserMedia() are applied.

[Radar WebKit Bug Importer]

<rdar://problem/29191384>

[Eric Carlson]

Created attachment 294366 [details]
Proposed patch.

[WebKit Commit Bot]

Attachment 294366 [details] did not pass style-queue:


ERROR: Source/WebCore/platform/mock/MockRealtimeAudioSource.cpp:47:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/WebCore/platform/mediastream/mac/MockRealtimeVideoSourceMac.mm:55:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/WebCore/platform/mock/MockRealtimeVideoSource.cpp:55:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 3 in 38 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Sam Weinig]

Comment on attachment 294366 [details]
Proposed patch.

View in context: https://bugs.webkit.org/attachment.cgi?id=294366&action=review

> Source/WebCore/platform/mediastream/MediaConstraints.h:279
> +            value = m_ideal.value();
> +            if (value < min)
> +                value = min;
> +            if (value > max)
> +                value = max;

You could write this as value = std::max(max, std::min(min,  m_ideal.value()));

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:168
> +        if (fitnessDistance(*widthConstraint) == std::numeric_limits<double>::infinity()) {

We traditionally use isinf) to compare against infinity.

[Eric Carlson]

Created attachment 294372 [details]
Updated patch.

[WebKit Commit Bot]

Attachment 294372 [details] did not pass style-queue:


ERROR: Source/WebCore/platform/mock/MockRealtimeAudioSource.cpp:47:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/WebCore/platform/mediastream/mac/MockRealtimeVideoSourceMac.mm:55:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/WebCore/platform/mock/MockRealtimeVideoSource.cpp:55:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 3 in 38 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Eric Carlson]

(In reply to comment #4)
> Comment on attachment 294366 [details]
> Proposed patch.
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=294366&action=review
> 
> > Source/WebCore/platform/mediastream/MediaConstraints.h:279
> > +            value = m_ideal.value();
> > +            if (value < min)
> > +                value = min;
> > +            if (value > max)
> > +                value = max;
> 
> You could write this as value = std::max(max, std::min(min, 
> m_ideal.value()));
> 
Fixed.

> > Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:168
> > +        if (fitnessDistance(*widthConstraint) == std::numeric_limits<double>::infinity()) {
> 
> We traditionally use isinf) to compare against infinity.
>
Fixed.

[Build Bot]

Comment on attachment 294372 [details]
Updated patch.

Attachment 294372 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/2490805

New failing tests:
fast/mediastream/apply-constraints-audio.html
fast/mediastream/apply-constraints-advanced.html
fast/mediastream/apply-constraints-video.html

[Build Bot]

Created attachment 294377 [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

[youenn fablet]

Comment on attachment 294372 [details]
Updated patch.

r=me once bots are all happy.
Some nits below.


View in context: https://bugs.webkit.org/attachment.cgi?id=294372&action=review

> Source/WebCore/platform/mediastream/CaptureDeviceManager.cpp:113
> +        if (RealtimeMediaSource* mediaSource = createMediaSourceForCaptureDeviceWithConstraints(captureDevice, constraints, invalidConstraint))

I would use auto for mediaSource.
It seems strange to retrieve a raw pointer also.
Why createMediaSourceForCaptureDeviceWithConstraints is not returning a RefPtr<>?

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:164
> +    RealtimeMediaSourceCapabilities& capabilities = *this->capabilities();

We probably should return a RealtimeMediaSourceCapabilities& directly.
Maybe add an ASSERT here in the meantime?

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:199
> +    // Each of the values is supported individually, see of they all can be applied at the same time.

s/of/if/

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:587
> +    {

Why do we need that indenting?

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:588
> +        RealtimeMediaSourceCapabilities& capabilities = *this->capabilities();

ASSERT again?

> Source/WebCore/platform/mediastream/RealtimeMediaSource.cpp:629
>  void RealtimeMediaSource::applyConstraints(const MediaConstraints& constraints, SuccessHandler successHandler, FailureHandler failureHandler)

It seems strange to pass callbacks.
I was expecting applyConstraints to work asynchronously, but it seems everything is done synchronously.
Can we simplify this?
It seems this function could just return Optional<String> or String if a null string means no error?

> Source/WebCore/platform/mediastream/openwebrtc/RealtimeMediaSourceCenterOwr.cpp:101
> +    // FIXME: Actually apply constraints to  newly created sources.

two spaces after to

[Eric Carlson]

Created attachment 294397 [details]
Patch for landing.

[WebKit Commit Bot]

Comment on attachment 294397 [details]
Patch for landing.

Clearing flags on attachment: 294397

Committed r208559: <http://trac.webkit.org/changeset/208559>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.