Fix two truncation bugs in harfbuzz-indic shaper.

Fix two truncation bugs in harfbuzz-indic shaper.

This bugs in isMark and isLetter lead 'init' feature to be applied
even when a syllable is not at the beginning of a word.

See http://www.microsoft.com/typography/otfntdev/bengalot/features.htm
for 'init' feature application in Bengali.

It turned out that the upstream also fixed this (see https://bugs.freedesktop.org/show_bug.cgi?id=30319) exactly the same way.

Change the upstream reference to 93d8573ddcf97648983a1c1b5cdda8a3ec6d4432
in README.chromium and remove 2 patches in chromium.patch because
they're all landed in the upstream @93d8573ddcf97648
Two removed patches are:

  - Fix OOB access (in Tibetan shaper) @ff0612c2e7df1b86fc
  - Fix crash! (in harfbuzz-gpos.c) @81c8ef785b079980a

In addition, the following upstream fix is included by going to 93d8573ddc.

  - Fix bug in contrib/harfbuzz-unicode.cc @93d8573ddcf9

Webkit bug to add a layout test for 'init' is
   https://bugs.webkit.org/show_bug.cgi?id=74310

BUG=107031
TEST=With Lohit Bengali font installed (the default on CrOS devices), go to http://i18nl10n.com/fonts/beng3.html and the result should be the same as the left in the screenshot attached to the bug instead of the right. A webkit layout test will be added and will be mentioned in the bug later.
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=114082

[jungshik at Google, 2011-12-10 00:22:07]

We took the upstream at d10a264823e (Jul 13, 2011) and the fix for this was
landed to 81f2ecafa19b602 on June 14, 2011. Apparently, the latter was not
merged to the main line when we took this out.

Now we're up to 93d8573ddcf9764 with harfbuzz-unicode.c change added. 


Do we want to keep three diffs in chrome.patch that are already included in the
upstream at 93d8573ddcf9764 or do we want to remove them and update
README.chromium to point to 93d8573ddcf9764 ?

[bashi, 2011-12-12 00:13:10]

LGTM.

I think we'd better update README.chromium to point to 93d8573ddcf9764 and
remove the diffs.

On 2011/12/10 00:22:07, Jungshik Shin wrote:
> We took the upstream at d10a264823e (Jul 13, 2011) and the fix for this was
> landed to 81f2ecafa19b602 on June 14, 2011. Apparently, the latter was not
> merged to the main line when we took this out.
> 
> Now we're up to 93d8573ddcf9764 with harfbuzz-unicode.c change added. 
> 
> 
> Do we want to keep three diffs in chrome.patch that are already included in
the
> upstream at 93d8573ddcf9764 or do we want to remove them and update
> README.chromium to point to 93d8573ddcf9764 ?

[behdad_google, 2011-12-12 21:12:25]

LGTM.

[jungshik at Google, 2011-12-12 21:36:43]

Thank you for the review. I updated README to point to the latest in the
upstream and removed diffs that are not necessary any more from chromium.patch.