Updates for the skimage tool.

Updates for the skimage tool.

Allow passing files or folders on the command line.

Group the output to show all successes together, all failures
together, etc.

When writing a new png, do not make its file type ".png.png"
if the original was a png.

Force linking for JPEG decoder.

Use SkCommandLineFlags.

Committed: https://code.google.com/p/skia/source/detail?r=8615

[djsollen, 2013-04-11 14:17:44]

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp
File tools/skimage_main.cpp (right):

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:50: return true;
add the srcPath to successes here so that all the operations are in the same
place.

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:87: void decodeFileAndWrite(const char filePath[], const
SkString* writePath);
why not just move the function here?

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:151: result.printf("%s [%d %d]", filePath,
bitmap.width(), bitmap.height());
move these two lines in decodeFile

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:161: void forceLinking();
this isn't needed.

[robertphillips, 2013-04-11 14:29:53]

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp
File tools/skimage_main.cpp (right):

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:161: void forceLinking();
I think it was added elsewhere to silence some iOS warning.

[scroggo, 2013-04-11 15:32:20]

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp
File tools/skimage_main.cpp (right):

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:50: return true;
On 2013/04/11 14:17:44, djsollen wrote:
> add the srcPath to successes here so that all the operations are in the same
> place.

Done. decodeAndWriteFile still adds a string stating where the output file was
drawn (which this function does not know). I could combine the two functions
completely though. Let me know if you think that is better.

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:87: void decodeFileAndWrite(const char filePath[], const
SkString* writePath);
On 2013/04/11 14:17:44, djsollen wrote:
> why not just move the function here?

I was hoping to make the diff simpler, but it comes at the expense of
readability. I'll move it.

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:151: result.printf("%s [%d %d]", filePath,
bitmap.width(), bitmap.height());
On 2013/04/11 14:17:44, djsollen wrote:
> move these two lines in decodeFile

Done.

https://codereview.chromium.org/14089002/diff/2001/tools/skimage_main.cpp#new...
tools/skimage_main.cpp:161: void forceLinking();
On 2013/04/11 14:29:53, robertphillips wrote:
> I think it was added elsewhere to silence some iOS warning.

Cary added these to fix warnings on Mac in
https://code.google.com/p/skia/source/detail?r=4185. I tested on Mac and iOS,
and I see the warning on Mac but not on iOS. Should the prototype only be used
on platforms that give a warning? (i.e. put it inside #ifdef BUILD_FOR_MAC)

[djsollen, 2013-04-11 15:40:16]

lgtm

[scroggo, 2013-04-11 15:53:44]

Committed patchset #3 manually as r8615 (presubmit successful).