Issue 1663323002: Fix Unpremul Index8 tests

CQ is trying da patch. Follow status at https://chromium-cq-status.appspot.com/patch-status/1663323002/1 View timeline at https://chromium-cq-status.appspot.com/patch-timeline/1663323002/1

4 years, 10 months ago (2016-02-03 23:03:28 UTC) #7

commit-bot: I haz the power

Committed patchset #1 (id:1) as https://skia.googlesource.com/skia/+/e1daa48b9a5f1724cbfa6b00ba964f7fdfe0c3d0

4 years, 10 months ago (2016-02-03 23:31:20 UTC) #9

scroggo

On 2016/02/03 23:02:36, msarett wrote: > Gold is a little bit crazy right now. Three ...

4 years, 10 months ago (2016-02-04 13:39:52 UTC) #10

msarett

4 years, 10 months ago (2016-02-04 14:59:11 UTC) #11

Message was sent while issue was closed.

On 2016/02/04 13:39:52, scroggo wrote:
> On 2016/02/03 23:02:36, msarett wrote:
> > Gold is a little bit crazy right now.  Three known issues:
> > (1) The BMP bug (which I think I've fixed).
> > (2) This bug.
> > (3) Unpremul WEBP - Looks like a Skia premultiply is different than a WEBP
> > premultiply.
> > 
> > I'm going to go ahead and land this.  And then hopefully tomorrow morning
the
> > only issue will be with the WEBPs.
> 
> Sgtm. It doesn't surprise me that WEBP premultiplies differently. (After all,
> Blink and libpng do.) It might also be interesting to know whether we can
> premultiply faster. On the other hand, we currently do not have to do any
> swizzling for WEBP, so premultiplying ourselves would be an extra step.
> 

Yes!  Will look to see if we can premultiply faster!

> But we do not insist that they be exactly the same, right? So I'm guessing we
> just have one more version for each scale/etc that looks slightly different?

That's correct.

Expand Messages | Collapse Messages | Show Generated Messages | Hide Generated Messages