Change the name of SkFlags to SkCommandLineFlags.

Change the name of SkFlags to SkCommandLineFlags.

This name is more specific to what it actually does.

Also move the code into tools/flags, to (slightly) better organize
the massive tools folder.

Update the programs that use it to use the new names.

No functionality changes.

BUG=https://code.google.com/p/skia/issues/detail?id=1173

Committed: https://code.google.com/p/skia/source/detail?r=8304

[reed1, 2013-03-21 14:30:11]

For the future, is it ever useful to not force this guy to use globals, but just
an instance? The caller can always have a global ptr themselves, and "point" to
the instance that they created...

https://codereview.chromium.org/12440067/diff/1/tools/flags/SkCommandLineFlags.h
File tools/flags/SkCommandLineFlags.h (right):

https://codereview.chromium.org/12440067/diff/1/tools/flags/SkCommandLineFlag...
tools/flags/SkCommandLineFlags.h:111: static void ParseCommandLine(int argc,
char** argv);
Can rename this to just "Parse" if you want, now that CommandLine is in the
class name.

[scroggo, 2013-03-21 19:04:15]

> For the future, is it ever useful to not force this guy to use globals, but
just
> an instance? The caller can always have a global ptr themselves, and "point"
to
> the instance that they created...

I think the simplicity offered by using globals, including the ability to access
from anywhere, is valuable. Since this is only used by our tools, I don't see
much danger from the globals.

If we change to using an instance, we'll also need to change the API for
accessing variables set by the flags. I'm fine with changing it if there's
something we like better, but I kind of like the current model:

if (FLAGS_boolean) {
     ...

To grab a flag from an object, the only way I can think to do it is something
like the following:

if (myFlagsInstance.getBoolean("boolean")) {
     ...

which I find to be quite a bit uglier. I suppose we could have the client create
a subclass (so the accessor would be something like
bench_pictureFlags.enable-missing-warning()), but that seems like it would get
messy too.

https://codereview.chromium.org/12440067/diff/1/tools/flags/SkCommandLineFlags.h
File tools/flags/SkCommandLineFlags.h (right):

https://codereview.chromium.org/12440067/diff/1/tools/flags/SkCommandLineFlag...
tools/flags/SkCommandLineFlags.h:111: static void ParseCommandLine(int argc,
char** argv);
On 2013/03/21 14:30:11, reed1 wrote:
> Can rename this to just "Parse" if you want, now that CommandLine is in the
> class name.

Done.

[reed1, 2013-03-21 19:14:20]

lgtm

[scroggo, 2013-03-21 19:43:28]

Committed patchset #2 manually as r8304 (presubmit successful).