[WebGPU] Move 'gpu' API entry point from DOMWindow to Navigator

[Justin Fan]

[WebGPU] Move 'gpu' API entry point from DOMWindow to Navigator

[Justin Fan]

Created attachment 368417 [details]
Patch

[Justin Fan]

Created attachment 368418 [details]
Patch

[EWS Watchlist]

Comment on attachment 368418 [details]
Patch

Attachment 368418 [details] did not pass mac-wk2-ews (mac-wk2):
Output: https://webkit-queues.webkit.org/results/12019461

New failing tests:
fast/dom/navigator-detached-no-crash.html

[EWS Watchlist]

Created attachment 368419 [details]
Archive of layout-test-results from ews104 for mac-highsierra-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews104  Port: mac-highsierra-wk2  Platform: Mac OS X 10.13.6

[Justin Fan]

Created attachment 368482 [details]
Patch

[Justin Fan]

Created attachment 368484 [details]
Patch

[Justin Fan]

Created attachment 368492 [details]
Patch

[Chris Dumez]

Comment on attachment 368492 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=368492&action=review

> Source/WebCore/ChangeLog:6
> +        Latest API provides 'gpu' through Navigator instead of DOMWindow. Replace DOMWindowWebGPU with NavigatorGPU.

Is this meant to be exposed to workers in any way?

> Source/WebCore/ChangeLog:8
> +        Reviewed by NOBODY (OOPS!).

This should be *before* the changelog description.

> Source/WebCore/Modules/webgpu/NavigatorGPU.idl:32
> +    [SameObject] readonly attribute WebGPU gpu;

Note that [SameObject] is currently a no-op AFAIK and you'll need to write custom bindings code (i.e. visitor) to ensure that the same object is always returned.

[Justin Fan]

(In reply to Chris Dumez from comment #8)
> Comment on attachment 368492 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=368492&action=review
> 
> > Source/WebCore/ChangeLog:6
> > +        Latest API provides 'gpu' through Navigator instead of DOMWindow. Replace DOMWindowWebGPU with NavigatorGPU.
> 
> Is this meant to be exposed to workers in any way?
> 
> > Source/WebCore/ChangeLog:8
> > +        Reviewed by NOBODY (OOPS!).
> 
> This should be *before* the changelog description.
> 
> > Source/WebCore/Modules/webgpu/NavigatorGPU.idl:32
> > +    [SameObject] readonly attribute WebGPU gpu;
> 
> Note that [SameObject] is currently a no-op AFAIK and you'll need to write
> custom bindings code (i.e. visitor) to ensure that the same object is always
> returned.

Thanks for looking. Yes, there's also a partial interface for WorkerNavigator; I'll update.

[EWS Watchlist]

Comment on attachment 368492 [details]
Patch

Attachment 368492 [details] did not pass mac-wk2-ews (mac-wk2):
Output: https://webkit-queues.webkit.org/results/12033795

New failing tests:
fast/dom/navigator-detached-no-crash.html

[EWS Watchlist]

Created attachment 368503 [details]
Archive of layout-test-results from ews107 for mac-highsierra-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews107  Port: mac-highsierra-wk2  Platform: Mac OS X 10.13.6

[Justin Fan]

Created attachment 368528 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 368528 [details]
Patch

Clearing flags on attachment: 368528

Committed r244777: <https://trac.webkit.org/changeset/244777>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/50337891>