Reland SkWriter32 growth change with build fixes.

Reland SkWriter32 growth change with build fixes.

  - SkSWriter32 resets itself with its stack block;
  - Track the full capacity of fInternal (which may be >size after a reset).

BUG=skia:2125
Committed: http://code.google.com/p/skia/source/detail?r=13412

[reed1, 2014-02-11 16:31:31]

seems fine -- lgtm -- does the recent snafu suggest any unittests we could write
for this aspect?

[mtklein, 2014-02-11 16:46:30]

On 2014/02/11 16:31:31, reed1 wrote:
> seems fine -- lgtm -- does the recent snafu suggest any unittests we could
write
> for this aspect?

Maybe.  The writer unit tests and GM caught the problem.  They weren't of course
specifically testing for this bug, but they do cover it.

I think what's suggested here is to:
  - clean up some pointlessly non-portable code in the writer tests that made
this harder to catch on 64-bit machines
  - keep pruning SkWriter32's API down
  - start running tests on the commit queue

[reed1, 2014-02-11 16:51:00]

On 2014/02/11 16:46:30, mtklein wrote:
> On 2014/02/11 16:31:31, reed1 wrote:
> > seems fine -- lgtm -- does the recent snafu suggest any unittests we could
> write
> > for this aspect?
> 
> Maybe.  The writer unit tests and GM caught the problem.  They weren't of
course
> specifically testing for this bug, but they do cover it.
> 
> I think what's suggested here is to:
>   - clean up some pointlessly non-portable code in the writer tests that made
> this harder to catch on 64-bit machines
>   - keep pruning SkWriter32's API down
>   - start running tests on the commit queue

Ah, great that it *was* caught by a unit test.

+1 for running tests *somewhere* in the CQ -- I expect they're pretty fast.

[mtklein, 2014-02-11 17:01:07]

On 2014/02/11 16:51:00, reed1 wrote:
> On 2014/02/11 16:46:30, mtklein wrote:
> > On 2014/02/11 16:31:31, reed1 wrote:
> > > seems fine -- lgtm -- does the recent snafu suggest any unittests we could
> > write
> > > for this aspect?
> > 
> > Maybe.  The writer unit tests and GM caught the problem.  They weren't of
> course
> > specifically testing for this bug, but they do cover it.
> > 
> > I think what's suggested here is to:
> >   - clean up some pointlessly non-portable code in the writer tests that
made
> > this harder to catch on 64-bit machines
> >   - keep pruning SkWriter32's API down
> >   - start running tests on the commit queue
> 
> Ah, great that it *was* caught by a unit test.
> 
> +1 for running tests *somewhere* in the CQ -- I expect they're pretty fast.

(This may amaze you, but DM's faster.)

[mtklein, 2014-02-11 17:13:37]

The CQ bit was checked by mtklein@google.com

[commit-bot: I haz the power, 2014-02-11 17:13:44]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/mtklein@chromium.org/158953003/50001

[commit-bot: I haz the power, 2014-02-11 18:22:07]

Change committed as 13412