706052 - Unwrap to dom::Element instead of nsIDOMElement in texSubImage2D/texImage2D

Status: RESOLVED DUPLICATE of bug 730554

(Core :: Graphics: CanvasWebGL, defect)

Description

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

After bug 704088, that's all we need in WebGLContext::{TexSubImage2D_dom,TexImage2D_dom}.

Comment 1

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Attachment: Unwrap to dom::Element instead of nsIDOMElement in texSubImage2D/texImage2D;

Comment 2

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 584403 [details] [diff] [review]
Unwrap to dom::Element instead of nsIDOMElement in texSubImage2D/texImage2D;

This looks ok, but we can speed this up even more by adding a dom::Element specialization of xpc_qsUnwrapArg, right?  Should be pretty easy...

Comment 3

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Doesn't bug 563659 make that hard? (There is a specialization for nsGenericElement, but that unwraps to nsIContent first, so I don't see how it would be much of an improvement.)

Comment 4

[Boris Zbarsky [:bzbarsky]]

> (There is a specialization for nsGenericElement, but that unwraps to nsIContent first,
> so I don't see how it would be much of an improvement.)

It's an improvement because unrwapping to nsIContent does the fast-path unwrap using castNativeFromWrapper instead of having to call QueryInterface (which is what unwrap to dom::Element has to do right now).

So adding a specialization for dom::Element that looks just like the nsGenericElement one would be a win.

Comment 5

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Attachment: Part b: dom::Element specialization

Ah, that was the missing bit. I did xpc_qsUnwrapThis as well while I was there. (Never mind the NS_SPECIALIZE_TEMPLATE bits, this is an old tree without bug 706010.)

Comment 6

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 584812 [details] [diff] [review]
Part b: dom::Element specialization

r=me

Comment 7

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Applying the two patches causes crashes in M1:

https://tbpl.mozilla.org/?tree=Try&rev=ef8775dc7503

And just the first gives timeouts:

https://tbpl.mozilla.org/?tree=Try&rev=09222a9f4307

Comment 8

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

I guess the timeouts were because of the "if (NS_FAILED(rv)) return JS_FALSE;", which triggered because we were no longer unwrapping to a scriptable interface.

Anyway, fixed in bug 730554.