386229 - Setting cursor with GIF, PNG or XBM doesn't work

Status: VERIFIED FIXED

(Core Graveyard :: Image: Painting, defect)

Description

[Alfred Kayser]

Attachment: FIx 1bit cursor images and remove non-Cairo stuff

See test: http://home.unet.nl/alfredkayser/cursor_test.html
The first two pointers work, but the third result in a black image.

Furthermore in Cairo the code to set cursor image can be optimized.

Asking for blocking1.9 as this is a regression from FF2.0 (caused by Cairo)

Comment 1

[Robert Longson [:longsonr]]

+  // By passing negative heigth we tell Windows that the image is top-down

s/heigth/height/

Comment 2

[tor]

Comment on attachment 270211 [details] [diff] [review]
FIx 1bit cursor images and remove non-Cairo stuff

sr=tor with longsonr's nit fixed.

Comment 3

[Alfred Kayser]

Comment on attachment 270211 [details] [diff] [review]
FIx 1bit cursor images and remove non-Cairo stuff

Asking for Approval1.9 as this is a regression from FF2.0 (caused by Cairo)

Comment 4

[Alfred Kayser]

Attachment: V2: spelling of heigth fixed

Comment 5

[Alfred Kayser]

Attachment: V3: Unbitrotted and updated to handle transparency really in the right way.

Tor, can you recheck this patch?

More images have now transparency issues, see also http://e-vertise.com/warpzilla/OS2_Pointer_Test.html for more test cases showing this.

This patch solves all these issues, including the 16bit issue addressed in bug 383668.

This also removes the copying and reversing of the image data, as this is not needed, just pass a negative height to DataToBitmap and Windows takes care of the rest. Also remove Win95x stuff as it is no longer supported anyway.

Comment 6

[tor]

Going by patch size (28K to 10K), it seems your patch has changed considerably.  Shouldn't you get review again?

Comment 7

[Alfred Kayser]

Comment on attachment 278039 [details] [diff] [review]
V3: Unbitrotted and updated to handle transparency really in the right way.

The big difference in size is caused by the non-cairo removal meanwhile. 
But, best to be safe than sorry: Ere, can you re-review this patch?

Thanks in advance

Comment 8

[Alfred Kayser]

Attachment: V4: Use PR_Free to match PR_Calloc

Comment 9

[Alfred Kayser]

Comment on attachment 280873 [details] [diff] [review]
V4: Use PR_Free to match PR_Calloc

Can you please re-review tis patch (because of other updates meanwhile, the size of this patch has been reduced from 29KB to only 10KB).

Comment 10

[Alfred Kayser]

Re-asking for Blocking 1.9, as the patch become seriously smaller, and without this fix, set cursor images (GIF, PNG, and XBM formats) doesn't work anymore as it did in FF2.0. So this really needs to be fixed for FF3.0/Gecko 1.9 to be for this functionality on par with FF2.0.

Comment 11

[Ere Maijala (slow)]

Comment on attachment 280873 [details] [diff] [review]
V4: Use PR_Free to match PR_Calloc

r=emaijala

Comment 12

[Stuart Parmenter]

i don't think this should block 1.9 but it has approval+ so check it in before we go to code red.

Comment 13

[Reed Loden [:reed]]

Checking in widget/src/windows/nsWindow.cpp;
/cvsroot/mozilla/widget/src/windows/nsWindow.cpp,v  <--  nsWindow.cpp
new revision: 3.710; previous revision: 3.709
done
Checking in widget/src/windows/nsWindow.h;
/cvsroot/mozilla/widget/src/windows/nsWindow.h,v  <--  nsWindow.h
new revision: 3.243; previous revision: 3.242
done