183427 - Re-enable GDI alphablending

Status: RESOLVED FIXED

(Core Graveyard :: Image: Painting, defect)

Description

[ArronM (:paper)]

Regression from Bug 157521 caused GDI alphablending to be always turned off.
Part of the patch in 157521 that did it:

-  if (gPlatform == VER_PLATFORM_WIN32_WINDOWS && mSizeImage >= 0xFF0000) 
-    mCanOptimize = PR_FALSE;
-  else if ((8 != mAlphaDepth) || CanAlphaBlend())
-    mCanOptimize = PR_TRUE;
+  if ((gPlatform == VER_PLATFORM_WIN32_WINDOWS && mSizeImage >= 0xFF0000) ||
(mAlphaDepth == 8) || (!CanAlphaBlend()) ){
+    return NS_OK;
+  }  

The || needs to be an &&.  We can't optimize only in the case where alphadepth
is 8 and !CanAlphaBlend (as well as that imagesize limit)

Comment 1

[ArronM (:paper)]

Attachment: The Patch

I've also moved the mSizeImage 0 check up, as recommended by kin in Bug 157521
Comment 4

Comment 2

[ArronM (:paper)]

Attachment: diff -uw, for easier reading

Comment 3

[ArronM (:paper)]

Comment on attachment 108188 [details] [diff] [review]
The Patch

dcone, since this bug was caused by your checkin, could you review?

Comment 4

[ArronM (:paper)]

Comment on attachment 108188 [details] [diff] [review]
The Patch

rods, kin, could your r/sr since you two r/sr'd the patch that caused this
regression?

It's an AND/OR logic-type patch, plus a combining of a if statement.. no real
knowledge of nsImageWin is needed.

Comment 5

[kinmoz]

Comment on attachment 108188 [details] [diff] [review]
The Patch

sr=[e-mail]

Comment 6

[Simon Montagu :smontagu]

Comment on attachment 108188 [details] [diff] [review]
The Patch

r=smontagu, and if you added some comments to the condition before checking in
I would be even happier.

Comment 7

[ArronM (:paper)]

checked in earlier today