PDF: add support for named destinations

PDF: add support for named destinations

Imported from https://codereview.appspot.com/7374052/ on behalf of dml@google.com

Closed in favor of https://codereview.chromium.org/12466008

[epoger, 2013-02-28 19:43:26]

I'm going to try to run this through one or two trybots before committing...

[epoger, 2013-02-28 20:15:25]

Trybot runs are:

http://70.32.156.53:10117/builders/Skia_Linux_NoGPU_Debug_32_Trybot/builds/82

http://70.32.156.53:10117/builders/Skia_Shuttle_Win7_Intel_Float_Debug_64_Try...
 (not started yet)

[epoger, 2013-02-28 20:34:50]

> http://70.32.156.53:10117/builders/Skia_Linux_NoGPU_Debug_32_Trybot/builds/82

This trybot failed the GenerateGMs step:
---- internal_links_comparison-serialize: 1001 (of 350000) differing pixels, max
per-channel mismatch R=0 G=0 B=255 A=0

Which, looking at the "-serialize" code path at
https://code.google.com/p/skia/source/browse/trunk/gm/gmmain.cpp?r=7885#1563 ,
seems to indicate that the output image differs depending on whether the
SkPicture was rendered directly or serialized/deserialized...

Leon, any thoughts on this?

[David Lattimore, 2013-02-28 20:57:08]

On 2013/02/28 20:34:50, epoger wrote:
> >
http://70.32.156.53:10117/builders/Skia_Linux_NoGPU_Debug_32_Trybot/builds/82
> 
> This trybot failed the GenerateGMs step:
> ---- internal_links_comparison-serialize: 1001 (of 350000) differing pixels,
max
> per-channel mismatch R=0 G=0 B=255 A=0
> 
> Which, looking at the "-serialize" code path at
> https://code.google.com/p/skia/source/browse/trunk/gm/gmmain.cpp?r=7885#1563 ,
> seems to indicate that the output image differs depending on whether the
> SkPicture was rendered directly or serialized/deserialized...
> 
> Leon, any thoughts on this?

Sorry, I was seeing that locally, but wrongly assumed it was due to a missing
baseline for the new test. Should have realized that couldn't be the case when
it was mentioned that the baselines were stored separately.

I'll see if I can figure out what's going on. One theory is that the drawPoints
call is losing the isNoDrawAnnotation. Not sure how, since I'm doing the same as
drawRect with regard to isNoDrawAnnotation. Although, if there's no GM for
SkAnnotateRectWithURL then perhaps it has the same problem as well, but was
never noticed.

[epoger, 2013-02-28 21:08:27]

On 2013/02/28 20:57:08, dml wrote:
> On 2013/02/28 20:34:50, epoger wrote:
> > >
> http://70.32.156.53:10117/builders/Skia_Linux_NoGPU_Debug_32_Trybot/builds/82
> > 
> > This trybot failed the GenerateGMs step:
> > ---- internal_links_comparison-serialize: 1001 (of 350000) differing pixels,
> max
> > per-channel mismatch R=0 G=0 B=255 A=0
> > 
> > Which, looking at the "-serialize" code path at
> > https://code.google.com/p/skia/source/browse/trunk/gm/gmmain.cpp?r=7885#1563
,
> > seems to indicate that the output image differs depending on whether the
> > SkPicture was rendered directly or serialized/deserialized...
> > 
> > Leon, any thoughts on this?
> 
> Sorry, I was seeing that locally, but wrongly assumed it was due to a missing
> baseline for the new test. Should have realized that couldn't be the case when
> it was mentioned that the baselines were stored separately.

Yes, in that case it is NOT comparing against a prerecorded image... it's
comparing against another image generated on the fly, but through a different
path.