Update harfbuzz-ng to 0.9.14 from 0.9.10

Update harfbuzz-ng to 0.9.14 from 0.9.10

Old revision: 34e6c3e3e452bdf6f93df565a70453a6e74d4c6e
New revision: f872a17462a75a3493623747bf3a3fbe54556c7b 

What's new:

- A lot of Indic shaping bugs as well as a work-around for bugs in
  FreeSans (Devanagari)
- New Win8-style Myanmar shaper
- New shaper for Tai Tham, Chan, New Tai Lue

See NEWS file for more details.

While I'm at this, I also 'fixed' README.chromium to make the copyright 
information show up in about:credits page. 

Two tests requiring rebaselines (khmer-selection and danda-space) are added to
the list of expected failure until they're rebaselined.

See also http://webkit.org/b/113450 

BUG=chromium:96143
TEST=See comment #5 in the bug.
TBR=cpu

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=191666

[jungshik at Google, 2013-03-25 18:22:28]

Forgot to add new files to the CL

[jungshik at Google, 2013-03-25 19:16:39]

Tony, can you review this ? 
Behdad (the maintainer of harfbuzz) made some changes and he believes that we'd
better roll to fix the bug (mentioned in the description) and support
Burmese/Myanmar among other things.

[tony, 2013-03-25 19:22:26]

LGTM, but make sure you get a green layout test try run (or at least matching
the chromium.webkit DEPS bots) before landing.

Also, after this lands, please watch the intl page cycler bots for changes in
memory use or cycle time.

https://codereview.chromium.org/12438036/diff/19001/third_party/harfbuzz-ng/R...
File third_party/harfbuzz-ng/README.chromium (right):

https://codereview.chromium.org/12438036/diff/19001/third_party/harfbuzz-ng/R...
third_party/harfbuzz-ng/README.chromium:24: In addition, all unneeded files from
the root directory and src/ directory (hb-icu-le, hb-ucdn, hb-old, scripts to
generate data tables, automake/conf files)
Nit: 80 cols

[jungshik at Google, 2013-03-26 23:10:47]

Thank you for the review. 

3 layout test failed. One of them is completely unrelated to harfbuzz
(video-controls-caption.html). 

khmer-selection.html : The intended test is passed (Khmer selection), but text
and image dumps are still different. Actually, DumpRenderTree does not seem to
have access to a Khmer font and what's in the image (expected and actual) are a
bunch of empty rectangles. Somehow their dimensions are different between
expected and actual leading to a failure. 

danda-space.html : Locally, it passes (again with a bunch of empty rectangles),
but it failed in layout test bots. 

What would be the best way to handle this (aside from fixing layout tests /
DumpRenderTree to use real fonts for both tests)? I'd like to get this in for
M27 (I missed the branch cut-off yesterday and I'm planning to merge pending the
approval). 


On 2013/03/25 19:22:26, tony wrote:
> LGTM, but make sure you get a green layout test try run (or at least matching
> the chromium.webkit DEPS bots) before landing.
> 
> Also, after this lands, please watch the intl page cycler bots for changes in
> memory use or cycle time.
> 
>
https://codereview.chromium.org/12438036/diff/19001/third_party/harfbuzz-ng/R...
> File third_party/harfbuzz-ng/README.chromium (right):
> 
>
https://codereview.chromium.org/12438036/diff/19001/third_party/harfbuzz-ng/R...
> third_party/harfbuzz-ng/README.chromium:24: In addition, all unneeded files
from
> the root directory and src/ directory (hb-icu-le, hb-ucdn, hb-old, scripts to
> generate data tables, automake/conf files)
> Nit: 80 cols

[tony, 2013-03-26 23:26:55]

I would update the patch to include
webkit/tools/layout_tests/test_expectations.txt and mark the 2 related tests as
failing.

We can then examine the failure in more detail on the chromium.webkit bots.  I
think we'll just need to rebaseline, but I trust the chromium.webkit bots more
than the try bots.

[jungshik at Google, 2013-03-28 20:21:48]

Thank you for the advice. I'll do what you suggested. 

I'll also update the list of fonts for DumpRenderTree. Alternatively, we can
check in fonts (we developed and released under Apache v2 license) to the Webkit
repo and use them as web fonts. That way, a lot of uncertainty and bogus test
results (as we saw in this bug) in the expected files can be taken care of. 

On 2013/03/26 23:26:55, tony wrote:
> I would update the patch to include
> webkit/tools/layout_tests/test_expectations.txt and mark the 2 related tests
as
> failing.
> 
> We can then examine the failure in more detail on the chromium.webkit bots.  I
> think we'll just need to rebaseline, but I trust the chromium.webkit bots more
> than the try bots.

[jungshik at Google, 2013-03-28 20:22:19]

https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests...
File webkit/tools/layout_tests/test_expectations.txt (right):

https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests...
webkit/tools/layout_tests/test_expectations.txt:38: crbug.com/224428 [Linux]
fast/text/international/khmer-selection.html [ Failure ]
Tony, 

Is the syntax above correct? What's the way to specify that this test will fail
only on Linux? The syntax seems to have changed over time. 

This test still fails in layout_bots while the one below does not show up in the
failure list anymore.

[tony, 2013-03-28 20:30:19]

On 2013/03/28 20:22:19, Jungshik Shin wrote:
>
https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests...
> File webkit/tools/layout_tests/test_expectations.txt (right):
> 
>
https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests...
> webkit/tools/layout_tests/test_expectations.txt:38: crbug.com/224428 [Linux]
> fast/text/international/khmer-selection.html [ Failure ]
> Tony, 
> 
> Is the syntax above correct? What's the way to specify that this test will
fail
> only on Linux? The syntax seems to have changed over time. 

The syntax looks correct to me and I tried it locally.  It doesn't work for me
either.  I tried to introduce a syntax error and --lint-test-files doesn't catch
it.

Dirk, do you know what's going on?  Maybe we should just suppress it upstream
and wait for WebKit to roll into Chromium.  I can do that for you.


Regarding fonts, I prefer we try to use fonts found on the system since that's
what most Linux users will use.

[jungshik at Google, 2013-03-28 22:28:44]

Thank you, Tony.

I've just landed an upstream CL to suppress those 2 tests, but made a
mistake of not updating LayoutTests/ChangeLog. Hope it'd not annoy webkit
folks.

Once Webkit is rolled to include that, I'll check in this CL without
test_expectation change (because it's not necessary).

Jungshik


2013/3/28 <tony@chromium.org>

> On 2013/03/28 20:22:19, Jungshik Shin wrote:
>
> https://codereview.chromium.**org/12438036/diff/43001/**
>
webkit/tools/layout_tests/**test_expectations.txt<https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests/test_expectations.txt>
>
>> File webkit/tools/layout_tests/**test_expectations.txt (right):
>>
>
>
> https://codereview.chromium.**org/12438036/diff/43001/**
>
webkit/tools/layout_tests/**test_expectations.txt#**newcode38<https://codereview.chromium.org/12438036/diff/43001/webkit/tools/layout_tests/test_expectations.txt#newcode38>
>
>> webkit/tools/layout_tests/**test_expectations.txt:38:
crbug.com/224428[Linux]
>> fast/text/international/khmer-**selection.html [ Failure ]
>> Tony,
>>
>
>  Is the syntax above correct? What's the way to specify that this test will
>>
> fail
>
>> only on Linux? The syntax seems to have changed over time.
>>
>
> The syntax looks correct to me and I tried it locally.  It doesn't work
> for me
> either.  I tried to introduce a syntax error and --lint-test-files doesn't
> catch
> it.
>
> Dirk, do you know what's going on?  Maybe we should just suppress it
> upstream
> and wait for WebKit to roll into Chromium.  I can do that for you.
>
>
> Regarding fonts, I prefer we try to use fonts found on the system since
> that's
> what most Linux users will use.
>
>
>
https://codereview.chromium.**org/12438036/<https://codereview.chromium.org/1...
>

[tony, 2013-03-28 23:15:13]

Sounds good to me. Thanks!

On 2013/03/28 22:28:44, Jungshik Shin wrote:
> Thank you, Tony.
> 
> I've just landed an upstream CL to suppress those 2 tests, but made a
> mistake of not updating LayoutTests/ChangeLog. Hope it'd not annoy webkit
> folks.
> 
> Once Webkit is rolled to include that, I'll check in this CL without
> test_expectation change (because it's not necessary).

[jungshik at Google, 2013-03-29 23:45:09]

Committing now that Webkit has rolled to include these two failures in
LayoutTests/platform/chromium/TestExpectations.

[commit-bot: I haz the power, 2013-03-29 23:46:06]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jshin@chromium.org/12438036/71001

[commit-bot: I haz the power, 2013-03-29 23:46:23]

Presubmit check for 12438036-71001 failed and returned exit status 1.

INFO:root:Found 52 file(s).

Running presubmit commit checks ...
Running /b/commit-queue/workdir/chromium/PRESUBMIT.py
Running /b/commit-queue/workdir/chromium/third_party/PRESUBMIT.py

** Presubmit Messages **
Your #include order seems to be broken. Send mail to
marja@chromium.org if this is not the case.
      third_party/harfbuzz-ng/src/hb-ot-shape-fallback.cc:28 \
      third_party/harfbuzz-ng/src/hb-shape-plan.cc:30 \
      third_party/harfbuzz-ng/src/test-buffer-serialize.cc:40

** Presubmit ERRORS **
Missing LGTM from an OWNER for these files:
    third_party/harfbuzz-ng/src/hb-ot-map-private.hh
    third_party/harfbuzz-ng/src/hb-ft.cc
    third_party/harfbuzz-ng/src/hb-ot-layout-private.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-thai.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic-machine.rl
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-arabic.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic.cc
    third_party/harfbuzz-ng/src/hb-shaper.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-arabic-fallback.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic-machine.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-myanmar.cc
    third_party/harfbuzz-ng/src/hb-version.h
    third_party/harfbuzz-ng/src/hb-common.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-private.hh
    third_party/harfbuzz-ng/src/hb-private.hh
    third_party/harfbuzz-ng/src/hb-fallback-shape.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic-private.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-indic-table.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-default.cc
    third_party/harfbuzz-ng/NEWS
    third_party/harfbuzz-ng/src/hb-ot-shape-fallback.cc
    third_party/harfbuzz-ng/src/hb-ot-layout-gsub-table.hh
    third_party/harfbuzz-ng/src/hb-warning.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-sea.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-normalize.cc
    third_party/harfbuzz-ng/src/hb-blob.h
    third_party/harfbuzz-ng/src/hb-ot-shape-fallback-private.hh
    third_party/harfbuzz-ng/src/test-buffer-serialize.cc
    third_party/harfbuzz-ng/src/hb-atomic-private.hh
    third_party/harfbuzz-ng/src/hb-buffer-private.hh
    third_party/harfbuzz-ng/src/hb-buffer.h
    third_party/harfbuzz-ng/src/hb-buffer.cc
    third_party/harfbuzz-ng/harfbuzz.gyp
    third_party/harfbuzz-ng/src/hb-mutex-private.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-myanmar-machine.hh
    third_party/harfbuzz-ng/src/hb-ot-shape.cc
    third_party/harfbuzz-ng/src/hb-ot-layout-gsubgpos-private.hh
    third_party/harfbuzz-ng/src/hb-buffer-deserialize-json.hh
    third_party/harfbuzz-ng/src/hb-shaper-private.hh
    third_party/harfbuzz-ng/src/hb-ot-shape-private.hh
    third_party/harfbuzz-ng/src/hb-shape.cc
    third_party/harfbuzz-ng/src/hb-ot-layout-gpos-table.hh
    third_party/harfbuzz-ng/src/hb-ot-map.cc
    third_party/harfbuzz-ng/src/hb-ot-layout.cc
    third_party/harfbuzz-ng/src/hb-open-type-private.hh
    third_party/harfbuzz-ng/src/hb-shape-plan.cc
    third_party/harfbuzz-ng/src/hb-ot-shape-complex-sea-machine.hh
    third_party/harfbuzz-ng/src/hb-unicode-private.hh
    third_party/harfbuzz-ng/README.chromium
    third_party/harfbuzz-ng/src/hb-buffer-deserialize-text.hh
    third_party/harfbuzz-ng/src/hb-blob.cc
    third_party/harfbuzz-ng/src/hb-buffer-serialize.cc

Presubmit checks took 3.2s to calculate.

[jungshik at Google, 2013-03-29 23:50:57]

@cpu, can you give us owner review (third_party/), Thanks.

[jungshik at Google, 2013-03-29 23:51:37]

@cpu, can you approve (third_party/ owner)? thanks

[commit-bot: I haz the power, 2013-04-01 18:07:26]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jshin@chromium.org/12438036/85001

[jungshik at Google, 2013-04-01 18:08:06]

TBR'ing because behdad is the lead developer of harfbuzz-ng and tony is familiar
with Webkit part that uses it.

[commit-bot: I haz the power, 2013-04-01 21:30:36]

Change committed as 191666

[bungeman-chromium, 2013-04-17 18:28:34]

https://chromiumcodereview.appspot.com/12438036/diff/85001/third_party/harfbu...
File third_party/harfbuzz-ng/harfbuzz.gyp (right):

https://chromiumcodereview.appspot.com/12438036/diff/85001/third_party/harfbu...
third_party/harfbuzz-ng/harfbuzz.gyp:68:
'src/hb-ot-shape-complex-indic-table.hh',
This (hb-ot-shape-complex-indic-table.hh) needs to be removed, as it was deleted
with
http://cgit.freedesktop.org/harfbuzz/commit/src/hb-ot-shape-complex-indic-tab...

(issue found with CMake gyp generator, see
https://codereview.chromium.org/12314014/ )