799465 - Add complete support for non-nsISupports objects in new DOM bindings

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Peter Van der Beken [:peterv]]

Two things still assume that the native is an nsISupports: parent wrapping and anything related to the CC (traversal of the binding and wrapper preservation).

Comment 1

[Peter Van der Beken [:peterv]]

Attachment: Fix CC traversal and wrapper preservation v1

Comment 2

[Peter Van der Beken [:peterv]]

Attachment: Allow non-nsISupports object to be parents v1

Comment 3

[Peter Van der Beken [:peterv]]

Attachment: Make AudioBuffer not inherit from nsISupports v1

This was my testcase for the other two patches.

Comment 4

[(no longer active)]

Comment on attachment 669522 [details] [diff] [review]
Make AudioBuffer not inherit from nsISupports v1

Review of attachment 669522 [details] [diff] [review]:
-----------------------------------------------------------------

Yay!

Comment 5

[(no longer active)]

Can I do this for the rest of the web audio objects as well?

Comment 6

[Peter Van der Beken [:peterv]]

Sure, if they only have nsISupports to deal with CC it'd be good to do that. If you run into any trouble let us know, but it should work with these patches.

Comment 7

[(no longer active)]

(In reply to comment #6)
> Sure, if they only have nsISupports to deal with CC it'd be good to do that. If
> you run into any trouble let us know, but it should work with these patches.

OK will do.  (I'll wait for this to land first)

Comment 8

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 669520 [details] [diff] [review]
Fix CC traversal and wrapper preservation v1

r=me assuming NoteNativeRoot is the right thing to call there at the end.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 669521 [details] [diff] [review]
Allow non-nsISupports object to be parents v1

>+inline T
>+GetParentPointer(T aObject)

I would prefer that as:

  inline T*
  GetParentPointer(T* aObject)

so we definitely get useful compile failures if a type that's not a pointer and not a ParentObject is passed to GetParentPointer.

r=me with that.  Very nice!

Comment 10

[Peter Van der Beken [:peterv]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/7e7fc42021c1
https://hg.mozilla.org/integration/mozilla-inbound/rev/7154061ddc00
https://hg.mozilla.org/integration/mozilla-inbound/rev/a484a3a904da
https://hg.mozilla.org/integration/mozilla-inbound/rev/3044539fec87

I ended up disabling CC support for 'owned' natives, there's no easy way to implement a CC participant for those right now. I'll file a follow-up bug for that.

Comment 11

[Ed Morley [:emorley]]

Push backed out for compilation failures:
https://tbpl.mozilla.org/?tree=Mozilla-Inbound&rev=2dbcd6d16b43

https://hg.mozilla.org/integration/mozilla-inbound/rev/5540b310d435

Comment 12

[Peter Van der Beken [:peterv]]

(In reply to Ed Morley [:edmorley UTC+1] from comment #11)
> Push backed out for compilation failures:

Not caused by these patches.

https://hg.mozilla.org/integration/mozilla-inbound/rev/c1bbabd184b3
https://hg.mozilla.org/integration/mozilla-inbound/rev/a1c63e595136
https://hg.mozilla.org/integration/mozilla-inbound/rev/b4ecedf655f2

Comment 13

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/c1bbabd184b3
https://hg.mozilla.org/mozilla-central/rev/a1c63e595136
https://hg.mozilla.org/mozilla-central/rev/b4ecedf655f2