[webkit-reviews] review requested: [Bug 4940] CG and ATSUI give different width to the same text : [Attachment 4020] patch to implement the rounding hacks in the ATSUI code path

[bugzilla-request-daemon at opendarwin.org]

opendarwin.org at mitzpettel.com has asked  for review:
Bug 4940: CG and ATSUI give different width to the same text
http://bugzilla.opendarwin.org/show_bug.cgi?id=4940

Attachment 4020: patch to implement the rounding hacks in the ATSUI code path
http://bugzilla.opendarwin.org/attachment.cgi?id=4020&action=edit

------- Additional Comments from opendarwin.org at mitzpettel.com
Revised patch addressing Darin's comments:

> ATSULayoutParameters should not be in the header

Removed (I based ATSULayoutParameters on CharacterWidthIterator, which
currently is in the header, apparently for no good reason).

> letter-spacing bits

Removed.

> Is it right to use a copy of an ATSUTextLayout rather than a pointer in
> ATSUTextLayoutParameters?

ATSUTextLayout is itself a pointer to an opaque struct.

> there's no reason to call NewATSUDirectLayoutOperationOverrideUPP at all.

Changed to use the procptr directly. Also changed to install the override only
if word rounding is required (since otherwise it doesn't do anything).

> I think that given that initializeATSULayoutParameters creates the layout, it

> should have a name more like "create" and there should be a corresponding
> "dispose" function.

Renamed to createATSULayoutParameters and added disposeATSULayoutParameters.
Note that they don't allocate/free the params structure itself.

Also corrected coding style mistakes pointed out.

> Too bad my synthesized bold/oblique patch isn't landed
> yet. Merging is going to be a pain.

Indeed...