Check for the existence of git before running commands

Check for the existence of git before running commands.

Checks in the gclient wrapper and when running any gclient git command.

TEST=none
BUG=none

Patch contributed by Nasser <nasser@codeaurora>

[M-A Ruel, 2010-01-07 20:50:36]

http://codereview.chromium.org/518064/diff/1/3
File gclient_scm.py (right):

http://codereview.chromium.org/518064/diff/1/3#newcode294
gclient_scm.py:294: raise gclient_utils.Error("git command '%s' failed to run.\
We use string catenation instead of backslashes.

http://codereview.chromium.org/518064/diff/1/3#newcode300
gclient_scm.py:300: return output.strip()
Fix alignment to +2

[Nasser Grainawi, 2010-01-07 22:37:33]

On 2010/01/07 20:50:36, Marc-Antoine Ruel wrote:
> http://codereview.chromium.org/518064/diff/1/3
> File gclient_scm.py (right):
> 
> http://codereview.chromium.org/518064/diff/1/3#newcode294
> gclient_scm.py:294: raise gclient_utils.Error("git command '%s' failed to
run.\
> We use string catenation instead of backslashes.
> 
> http://codereview.chromium.org/518064/diff/1/3#newcode300
> gclient_scm.py:300: return output.strip()
> Fix alignment to +2

I cleared up the whitespace issues. Can you re-review?

[M-A Ruel, 2010-01-07 23:32:15]

lgtm, will commit later.

[M-A Ruel, 2010-01-08 01:12:43]

http://codereview.chromium.org/518064/diff/4001/4002
File gclient (right):

http://codereview.chromium.org/518064/diff/4001/4002#newcode28
gclient:28: echo -n "Trying git-svn... "
Actually, after trying the patch, I don't want this output nor line 12.

This is because it is always output otherwise. I'll commit without those lines.

[Lei Zhang, 2010-01-08 22:20:09]

So I'm only using gclient with svn, but with this CL, it complains about my git
version being too old. I think you need better detection to see if the user is
actually using git with gclient?

[Lei Zhang, 2010-01-08 22:23:51]

It looks like it's because I have depot_tools/git-cl-repo, even though I'm not
actively using it.

[M-A Ruel, 2010-01-08 22:26:08]

On 2010/01/08 22:23:51, Lei Zhang wrote:
> It looks like it's because I have depot_tools/git-cl-repo, even though I'm not
> actively using it.

I kind of agree the check is overly aggressive (in addition of slowing down
gclient). Nasser what do you think?

[Nasser Grainawi, 2010-01-08 22:29:29]

On 2010/01/08 22:26:08, Marc-Antoine Ruel wrote:
> On 2010/01/08 22:23:51, Lei Zhang wrote:
> > It looks like it's because I have depot_tools/git-cl-repo, even though I'm
not
> > actively using it.
> 
> I kind of agree the check is overly aggressive (in addition of slowing down
> gclient). Nasser what do you think?

We can drop the check in gclient. The one in gclient_scm.py should catch
anything real (without slowing anyone down). I'll send in a new patch or you can
just revert all the changes to gclient.