Port SkBase64 test to our test driver.

Port SkBase64 test to our test driver.

BUG=None
TESTS=tests --match SkBase64Test
R=mtklein@google.com, reed@google.com

Committed: https://code.google.com/p/skia/source/detail?r=13254

[tfarina, 2014-01-30 01:17:51]

Is that ok?

[mtklein, 2014-01-30 02:33:02]

Bunch of little suggestions.

https://codereview.chromium.org/132233060/diff/1/gyp/tests.gyp
File gyp/tests.gyp (right):

https://codereview.chromium.org/132233060/diff/1/gyp/tests.gyp#newcode142
gyp/tests.gyp:142: '../tests/SkBase64Test.cpp',
We have an odd precedent of leaving the Sk off the front of test files and
names.  This is just FYI.  I rather prefer the Sk prefix, but you may want to
drop it if you want consistency.

https://codereview.chromium.org/132233060/diff/1/src/utils/SkBase64.h
File src/utils/SkBase64.h (right):

https://codereview.chromium.org/132233060/diff/1/src/utils/SkBase64.h#newcode27
src/utils/SkBase64.h:27: size_t length() const { return fLength; }
Let's drop this and remove that assert in the test.  There's not much point
adding to the public API just to test it.  (If we ever find the need to call
length() outside the tests, we'll put it back.)

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp
File tests/SkBase64Test.cpp (right):

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:13: signed char all[256];
While we're at it, there are a few things we can clean up here:

 - signed char -> char.  signed char is sort of unnecessarily picky

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:15: all[index] = (signed char) (index + 1);
Let's add {} around this.

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:20: char* src = (char*)sk_malloc_throw(encodeLength + 1);
Can you make this guy an SkAutoMalloc or SkAutoTMalloc<char> ?  That'll
eliminate the need to pair it up with a free later.

[tfarina, 2014-01-30 02:44:59]

https://codereview.chromium.org/132233060/diff/1/src/utils/SkBase64.h
File src/utils/SkBase64.h (right):

https://codereview.chromium.org/132233060/diff/1/src/utils/SkBase64.h#newcode27
src/utils/SkBase64.h:27: size_t length() const { return fLength; }
On 2014/01/30 02:33:02, mtklein wrote:
> Let's drop this and remove that assert in the test.  There's not much point
> adding to the public API just to test it.  (If we ever find the need to call
> length() outside the tests, we'll put it back.)

Done.

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp
File tests/SkBase64Test.cpp (right):

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:13: signed char all[256];
On 2014/01/30 02:33:02, mtklein wrote:
> While we're at it, there are a few things we can clean up here:
> 
>  - signed char -> char.  signed char is sort of unnecessarily picky

Done.

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:15: all[index] = (signed char) (index + 1);
On 2014/01/30 02:33:02, mtklein wrote:
> Let's add {} around this.

Done.

https://codereview.chromium.org/132233060/diff/1/tests/SkBase64Test.cpp#newco...
tests/SkBase64Test.cpp:20: char* src = (char*)sk_malloc_throw(encodeLength + 1);
On 2014/01/30 02:33:02, mtklein wrote:
> Can you make this guy an SkAutoMalloc or SkAutoTMalloc<char> ?  That'll
> eliminate the need to pair it up with a free later.

Done.

[mtklein, 2014-01-30 02:50:30]

lgtm

[tfarina, 2014-01-30 22:51:53]

Committed patchset #2 manually as r13254 (presubmit successful).