[HarfBuzz] Refactoring mark skipping

[Jonathan Kew]

Still not quite right, according to my testing in Firefox - it now shapes the Lam of Scheherazade's lam-alef properly, but does not get the correct form for the Alef, because the backward-skipper fails.

My proposed fix (with a couple of bonus assertions, to check that we don't move the iterators further than they were initialized for) is attached; please check.

JK

-------------- next part --------------
A non-text attachment was scrubbed...
Name: mark-skipping-fixes
Type: application/octet-stream
Size: 1784 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20120118/068878d5/attachment.obj>
-------------- next part --------------


On 17 Jan 2012, at 23:09, Behdad Esfahbod wrote:

> On 01/17/2012 05:11 PM, Khaled Hosny wrote:
>> On Mon, Jan 16, 2012 at 10:22:40PM -0500, Behdad Esfahbod wrote:
>>> Hi Jonathan,
>>> 
>>> I refactored mark skipping all in one place and pushed to master.  Can you
>>> please review?  Also interested to see test results in Firefox.
>> 
>> Not sure if this is related, but now it seems some OpenType rules are not
>> triggered for the first and last last glyphs in a run, for example,
>> using Scheherazade '??' gives plain initial and final glyphs, while ' ?? '
>> give the correct contextual forms:
>> 
>>  $ hb-shape --no-positions --no-clusters ScheherazadeRegOT.ttf '??'
>>  <uni0627.fina|uni0644.init>
>> 
>>  $ hb-shape --no-positions --no-clusters ScheherazadeRegOT.ttf ' ?? '
>>  <space|uni0627.fina.postLamIni|uni0644.init.preAlef|space>
>> 
>> This was fine last week.
> 
> Fixed.
> 
> behdad
> 
> 
>> Regards,
>> Khaled
>>