Cursor support including custom cursors

Cursor support including custom cursors.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=5634

[Avi (use Gerrit), 2008-11-17 23:13:54]

Cursors!  (Random test page for that is
http://www.fontstuff.com/frontpage/fptut15.htm)

[darin (slow to review), 2008-11-18 01:43:49]

http://codereview.chromium.org/11427/diff/14/18
File webkit/glue/webcursor.cc (right):

http://codereview.chromium.org/11427/diff/14/18#newcode90
Line 90: #if defined(OS_MACOSX)
can we put the platform specific code into the platform specific files?  we have
webcursor_{port}.cc for this reason.

i also do not understand why we are using a CG bitmap.  why can't we use
SkBitmap on mac?

[Amanda Walker, 2008-11-18 02:52:28]

http://codereview.chromium.org/11427/diff/14/18
File webkit/glue/webcursor.cc (right):

http://codereview.chromium.org/11427/diff/14/18#newcode90
Line 90: #if defined(OS_MACOSX)
We'll be able to use SkBitmap after some other patches  (which are not yet
complete) that have the effect of changing the NativeImagePtr typedef to be an
SkBitmap (which it is not currently, since PLATFORM(CG) supercedes
PLATFORM(SKIA0 in the header files at the moment.

[Avi (use Gerrit), 2008-11-18 17:35:58]

For your consideration.

[darin (slow to review), 2008-11-18 18:16:27]

I didn't review webcursor_mac.mm, but the rest LG.

http://codereview.chromium.org/11427/diff/33/211
File webkit/glue/webcursor.h (right):

http://codereview.chromium.org/11427/diff/33/211#newcode79
Line 79: NSCursor* GetCursor() const;
Might be good to document the ownership model here.  Is this a NSCursor object
that the caller is responsible for destroying?

[pink (ping after 24hrs), 2008-11-18 18:28:12]

http://codereview.chromium.org/11427/diff/33/211
File webkit/glue/webcursor.h (right):

http://codereview.chromium.org/11427/diff/33/211#newcode79
Line 79: NSCursor* GetCursor() const;
On 2008/11/18 18:16:27, darin wrote:
> Might be good to document the ownership model here.  Is this a NSCursor object
> that the caller is responsible for destroying?

The usual nomenclature for Foundation/ObjC is that "getters" return an
autoreleased object, where as "create" return an object you're responsible for
releasing yourself. We can document it, but it's in line with expectations from
anyone using the code who knows the language.

[pink (ping after 24hrs), 2008-11-18 18:45:56]

http://codereview.chromium.org/11427/diff/33/209
File webkit/glue/webcursor_mac.mm (right):

http://codereview.chromium.org/11427/diff/33/209#newcode66
Line 66: return LoadCursor("crossHairCursor", 11, 11);
are these hotspot points documented anywhere, or are they just guesses from what
feels right using them? Are they in flipped coordindates or cocoa coordinates?

[Avi (use Gerrit), 2008-11-18 18:59:38]

http://codereview.chromium.org/11427/diff/33/209
File webkit/glue/webcursor_mac.mm (right):

http://codereview.chromium.org/11427/diff/33/209#newcode66
Line 66: return LoadCursor("crossHairCursor", 11, 11);
On 2008/11/18 18:45:56, pink wrote:
> are these hotspot points documented anywhere

See platform/mac/CursorMac.mm, as noted in the comment above. We're pulling
images from png files and there's no good way to note hotspots in the images
themselves. :(

> Are they in flipped coordindates or cocoa coordinates?

All NSCursor coordinates are flipped. See NSCursor documentation.

[pink (ping after 24hrs), 2008-11-18 20:52:48]

LGTM