Make running gn format unittests from anywhere work

Make running gn format unittests from anywhere work

This "works" but has two drawbacks:
- Relies on a 2-deep output directory which is true for gyp, but not
  necessarily for gn (though I'm pretty sure other things are broken
  for that case too).
- calls SetCurrentDirectory which is kind of icky. At first I tried setting
  an absolute path for loading the .golden file which was OK, but then
  it got messy to create the //gn path.

R=dpranke@chromium.org
BUG=443813
Committed: https://crrev.com/7f1ee75851d2a0f19ce8fe993cd7e9e1ccfebf4e
Cr-Commit-Position: refs/heads/master@{#309277}

[scottmg, 2014-12-19 03:51:23]

scottmg@chromium.org changed reviewers:
+ dpranke@chromium.org

[scottmg, 2014-12-19 04:03:02]

Better suggestions?

[Dirk Pranke, 2014-12-19 18:24:39]

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
File tools/gn/command_format_unittest.cc (right):

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
tools/gn/command_format_unittest.cc:25: PathService::Get(base::FILE_EXE, &exe); 
                               \
Can you use PathService::Get(base::DIR_SOURCE_ROOT) here instead? I think that's
the normally recommended mechanism for finding paths for test data.

[tfarina, 2014-12-19 18:29:04]

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
File tools/gn/command_format_unittest.cc (right):

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
tools/gn/command_format_unittest.cc:25: PathService::Get(base::FILE_EXE, &exe); 
                               \
On 2014/12/19 18:24:39, Dirk Pranke wrote:
> Can you use PathService::Get(base::DIR_SOURCE_ROOT) here instead? I think
that's
> the normally recommended mechanism for finding paths for test data.
+1 for that. I found DIR_SOURCE_ROOT easier to understand.

[scottmg, 2014-12-19 18:31:45]

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
File tools/gn/command_format_unittest.cc (right):

https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
tools/gn/command_format_unittest.cc:25: PathService::Get(base::FILE_EXE, &exe); 
                               \
On 2014/12/19 18:24:39, Dirk Pranke wrote:
> Can you use PathService::Get(base::DIR_SOURCE_ROOT) here instead? I think
that's
> the normally recommended mechanism for finding paths for test data.

Ah, indeed. Done.

(Sadly, it's doing the same thing, so is broken in the same way. But at least
it's broken the same way everywhere this way.)

[Dirk Pranke, 2014-12-19 20:23:18]

On 2014/12/19 18:31:45, scottmg wrote:
>
https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
> File tools/gn/command_format_unittest.cc (right):
> 
>
https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
> tools/gn/command_format_unittest.cc:25: PathService::Get(base::FILE_EXE,
&exe); 
>                                \
> On 2014/12/19 18:24:39, Dirk Pranke wrote:
> > Can you use PathService::Get(base::DIR_SOURCE_ROOT) here instead? I think
> that's
> > the normally recommended mechanism for finding paths for test data.
> 
> Ah, indeed. Done.
> 
> (Sadly, it's doing the same thing, so is broken in the same way. But at least
> it's broken the same way everywhere this way.)

lgtm. I'm don't think it's worth trying to do something more robust in this
situation, and it's hard
to say what you could really do that would be significantly better anyway.

(An alternative would be to compile the test data and results into the binary,
of course, but it's probably not worth doing that).

[scottmg, 2014-12-19 21:26:12]

On 2014/12/19 20:23:18, Dirk Pranke wrote:
> On 2014/12/19 18:31:45, scottmg wrote:
> >
>
https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
> > File tools/gn/command_format_unittest.cc (right):
> > 
> >
>
https://codereview.chromium.org/818623002/diff/1/tools/gn/command_format_unit...
> > tools/gn/command_format_unittest.cc:25: PathService::Get(base::FILE_EXE,
> &exe); 
> >                                \
> > On 2014/12/19 18:24:39, Dirk Pranke wrote:
> > > Can you use PathService::Get(base::DIR_SOURCE_ROOT) here instead? I think
> > that's
> > > the normally recommended mechanism for finding paths for test data.
> > 
> > Ah, indeed. Done.
> > 
> > (Sadly, it's doing the same thing, so is broken in the same way. But at
least
> > it's broken the same way everywhere this way.)
> 
> lgtm. I'm don't think it's worth trying to do something more robust in this
> situation, and it's hard
> to say what you could really do that would be significantly better anyway.
> 
> (An alternative would be to compile the test data and results into the binary,
> of course, but it's probably not worth doing that).

Yeah, I was going to do that and make them "regular" tests, but there's a lot of
fiddlyness with lines that are exactly the right length, quotes, etc. so they
get ugly/hard to read. Embedding as a bin2c or something seems like it'd be
annoying to debug. So I'll leave it like this for now.

[scottmg, 2014-12-19 21:26:15]

The CQ bit was checked by scottmg@chromium.org

[commit-bot: I haz the power, 2014-12-19 21:27:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/818623002/20001

[commit-bot: I haz the power, 2014-12-19 22:11:44]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2014-12-19 22:12:47]

Patchset 2 (id:??) landed as
https://crrev.com/7f1ee75851d2a0f19ce8fe993cd7e9e1ccfebf4e
Cr-Commit-Position: refs/heads/master@{#309277}