Make sure warnings fail the check when may_prompt=False.

Make sure warnings fail the check when may_prompt=False.

Split presubmit_support.Main() in two to make it simpler.

Handle better the case with unversioned files and add a unit test.

R=dpranke@chromium.org

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78700

[M-A Ruel, 2011-03-17 14:33:01]

Running presubmit hooks...
************* Module owners
R0201: 93:Database._check_collection: Method could be a function
** Presubmit Warnings **
Fix pylint errors first.

:)

[Dirk Pranke, 2011-03-17 19:12:56]

http://codereview.chromium.org/6674059/diff/1/presubmit_support.py
File presubmit_support.py (right):

http://codereview.chromium.org/6674059/diff/1/presubmit_support.py#newcode1155
presubmit_support.py:1155: 
Can we put this in scm.py? It doesn't feel like it belongs here.

http://codereview.chromium.org/6674059/diff/1/tests/presubmit_unittest.py
File tests/presubmit_unittest.py (right):

http://codereview.chromium.org/6674059/diff/1/tests/presubmit_unittest.py#new...
tests/presubmit_unittest.py:719: self.assertEquals(2, e.code)
This is the sort of thing that makes mock testing really painful for me. The
test is so white-box it's translucent.

The way I'd write this test if I had the tools I do in webkit (like
filesystem_mock) and the code structured the way I'd like would be to pass in a
fake filesystem and a fake subprocess wrapper, and then simply test the result
of the call (and possible that something was written to stderr, but not test
what exactly). As it is, this test is incredibly fragile; change almost anything
in the implementation, and the test breaks. Oh well. No need to change this now.

[M-A Ruel, 2011-03-18 15:54:21]

Updated.

[Dirk Pranke, 2011-03-18 16:16:04]

LGTM.

http://codereview.chromium.org/6674059/diff/6001/presubmit_support.py
File presubmit_support.py (right):

http://codereview.chromium.org/6674059/diff/6001/presubmit_support.py#newcode...
presubmit_support.py:1128: change_scm = scm.determine_scm(options.root)
At some point it seems like it might be better to change this to just be a
factory routine that returns an object of the right class.

[M-A Ruel, 2011-03-18 16:18:47]

On 2011/03/18 16:16:04, dpranke wrote:
>
http://codereview.chromium.org/6674059/diff/6001/presubmit_support.py#newcode...
> presubmit_support.py:1128: change_scm = scm.determine_scm(options.root)
> At some point it seems like it might be better to change this to just be a
> factory routine that returns an object of the right class.

Agreed, but the class is not in the right file at the moment and there's some
issue about "what's a change".