[webkit-reviews] review denied: [Bug 74741] Postpone deleteRenderbuffer/deleteTexture until all framebuffer attachment points are removed. : [Attachment 119655] Patch

[bugzilla-daemon at webkit.org]

Kenneth Russell <kbr at google.com> has denied Zhenyao Mo <zmo at google.com>'s
request for review:
Bug 74741: Postpone deleteRenderbuffer/deleteTexture until all framebuffer
attachment points are removed.
https://bugs.webkit.org/show_bug.cgi?id=74741

Attachment 119655: Patch
https://bugs.webkit.org/attachment.cgi?id=119655&action=review

------- Additional Comments from Kenneth Russell <kbr at google.com>
View in context: https://bugs.webkit.org/attachment.cgi?id=119655&action=review


> Source/WebCore/html/canvas/WebGLFramebuffer.cpp:245
> +	       gc3d->framebufferRenderbuffer(GraphicsContext3D::FRAMEBUFFER,
GraphicsContext3D::COLOR_ATTACHMENT0, GraphicsContext3D::RENDERBUFFER, 0);

These calls assume that this framebuffer object is bound to the FRAMEBUFFER
binding point at the time they're called. This may be guaranteed when the
WebGLRenderingContext calls the setAttachment method, but not when
deleteRenderbuffer, deleteTexture, etc. is called -- and these methods seem to
be the ones that call removeAttachment(WebGLObject*).

I don't understand the logic here regardless. I thought the desired behavior
was to defer the deletion of the WebGLTexture, WebGLRenderbuffer, etc. until it
was detached from all live framebuffers?