Bug 201333 – Introduce WorkerMessagePortChannelRegistry

• Component: Page Loading
• Status: RESOLVED FIXED

[youenn fablet]

Introduce WorkerMessagePortChannelRegistry

[youenn fablet]

Created attachment 377689 [details]
Patch

[youenn fablet]

Created attachment 377690 [details]
Patch

[Alex Christensen]

Comment on attachment 377690 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=377690&action=review

> Source/WebCore/dom/MessagePort.cpp:252
> +        auto scopeExit = makeScopeExit([&completionCallback] {
> +            completionCallback();
> +        });

Why make a lambda that calls the Function instead of just using the Function itself?  Why no WTFMove? Also, this seems like it should be a CompletionHandler.

> Source/WebCore/dom/MessagePort.cpp:315
> +        callOnMainThread([remoteIdentifier = m_remoteIdentifier, weakThis = makeWeakPtr(const_cast<MessagePort*>(this)), workerThread = WTFMove(workerThread)]() mutable {

makeWeakPtr should take const T& instead of non-const T& and this const_cast shouldn't be necessary. Same with const T*

[youenn fablet]

Created attachment 377829 [details]
Patch

[youenn fablet]

Created attachment 377839 [details]
Patch for landing

[youenn fablet]

(In reply to Alex Christensen from comment #3)
> Comment on attachment 377690 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=377690&action=review
> 
> > Source/WebCore/dom/MessagePort.cpp:252
> > +        auto scopeExit = makeScopeExit([&completionCallback] {
> > +            completionCallback();
> > +        });
> 
> Why make a lambda that calls the Function instead of just using the Function
> itself?  Why no WTFMove? Also, this seems like it should be a
> CompletionHandler.

Right.

> > Source/WebCore/dom/MessagePort.cpp:315
> > +        callOnMainThread([remoteIdentifier = m_remoteIdentifier, weakThis = makeWeakPtr(const_cast<MessagePort*>(this)), workerThread = WTFMove(workerThread)]() mutable {
> 
> makeWeakPtr should take const T& instead of non-const T& and this const_cast
> shouldn't be necessary. Same with const T*

makeWeakPtr can take a const pointer as input but it will create a WeakPtr of a const pointer.
In that case, we want to have a non const weakThis to call updateActivity.

[WebKit Commit Bot]

Comment on attachment 377839 [details]
Patch for landing

Clearing flags on attachment: 377839

Committed r249378: <https://trac.webkit.org/changeset/249378>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/54945252>