Update Android's FontHost to return NULL if familyName does not match

Update Android's FontHost to return NULL if familyName does not match

R=scroggo@google.com, wangxianzhu@chromium.org

Committed: https://code.google.com/p/skia/source/detail?r=11377

[djsollen, 2013-09-18 19:45:18]

If we land this change we will also need to land a change in chromium to take
advantage of this updated behavior.

[reed1, 2013-09-18 19:51:36]

https://codereview.chromium.org/23601041/diff/1/src/ports/SkFontConfigInterfa...
File src/ports/SkFontConfigInterface_android.cpp (right):

https://codereview.chromium.org/23601041/diff/1/src/ports/SkFontConfigInterfa...
src/ports/SkFontConfigInterface_android.cpp:496: SkASSERT(face);
Why are we asserting here?

[djsollen, 2013-09-18 19:57:30]

https://codereview.chromium.org/23601041/diff/1/src/ports/SkFontConfigInterfa...
File src/ports/SkFontConfigInterface_android.cpp (right):

https://codereview.chromium.org/23601041/diff/1/src/ports/SkFontConfigInterfa...
src/ports/SkFontConfigInterface_android.cpp:496: SkASSERT(face);
On 2013/09/18 19:51:37, reed1 wrote:
> Why are we asserting here?

This function takes fontRecID's which correspond to known fonts on the device
and creates a SkTypeface object around it.  Since we know the font exists and
what the name is then we expect it to succeed.  That being said we probably
don't need the assert here since it also exists later on in the function.

[scroggo, 2013-09-18 20:18:41]

lgtm

https://codereview.chromium.org/23601041/diff/1001/tests/Typeface.cpp
File tests/Typeface.cpp (right):

https://codereview.chromium.org/23601041/diff/1001/tests/Typeface.cpp#newcode24
tests/Typeface.cpp:24: SkAutoTUnref<SkTypeface>
t3(SkTypeface::CreateFromName("non-existent-font", SkTypeface::kNormal));
nit: 100 chars.

[Xianzhu, 2013-09-18 20:28:17]

Thanks Derek for the CL.

A question: I remember that Android framework needs the fallback logic. Wouldn't
this CL break Android framework?

If this behavior is for blink only, I think we can put BUILD_FOR_ANDROID_NDK
guard on the changed code.

[djsollen, 2013-09-18 20:30:15]

Yes, it does require a framework change and I've already uploaded that for
review (I'll add you to the CL).  I also met with the framework font team to
discuss this and feel that this is the best solution.

[Xianzhu, 2013-09-18 20:35:29]

Never mind. Just saw your CL in Android corresponding to this change.

[Xianzhu, 2013-09-18 20:35:57]

lgtm

[djsollen, 2013-09-19 12:08:48]

Committed patchset #2 manually as r11377 (presubmit successful).

[Xianzhu, 2013-10-04 16:29:53]

@djsollen, could you help close
https://code.google.com/p/skia/issues/detail?id=913 which has been fixed by this
CL? Thanks.