[Webkit-unassigned] [Bug 82956] [chromium] TransformationMatrix::to2dTransform seems unnecessary in LayerRendererChromium

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=82956





--- Comment #2 from Adrienne Walker <enne at google.com>  2012-04-02 16:17:29 PST ---
(In reply to comment #1)

> > As it stands, the function seems to be a gigantic no-op then, isn't it?  why do we need to do this operation?  It doesn't make anything faster; using 2d quads and points will already bypass the 3rd row and 3rd column, and everything else (inversion in particular) still operates on a 4x4 matrix.
> 
> The function is not a no-op.  For some values of transform,
> 
> transform.to2dtransform().inverse().mapQuad(quad) != transform.inverse().mapQuad(quad)

Haha, I realize this is exactly the same as the first thing you quote from me.  I think maybe I had some thought that not dropping the 3d transform part of the matrix would result in a different value when you projected back since you'd flattened that point to 2d and inflated in 2d.

Tests would certainly be helpful at this point in the conversation.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.