Add a --snapshot option for revinfo. If set, revinfo prints a .gclient file...

Add a --snapshot option for revinfo.  If set, revinfo prints a .gclient file
set to sync to the current revisions/hashes on disk

Add a --gclientfile=FILENAME option.  If set, gclient loads FILENAME
rather than .gclient for configuration info.

Patch contributed by jneddo@nvidia.com

BUG=1658
TEST=Verify that --snapshot prints a valid configuration file
     Verify that when --gclientfile is set the correct file is used
     Verify that gclient sync --gclientfile sets the correct revisions

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=41078

[jneddo1, 2010-03-08 20:43:55]

Adds a way to take a snapshot of the current tree

[M-A Ruel, 2010-03-09 15:19:23]

Sorry for the delay;

I'm fine with the idea, I'm just warry of having two similar functions diverging
even more when in fact they should be merged, or at least common path be
refactored in separate functions. Do you mind doing that?

http://codereview.chromium.org/672004/diff/1/2
File gclient.py (right):

http://codereview.chromium.org/672004/diff/1/2#newcode847
gclient.py:847: solution_names = {}
What about (from your code):
solutions_names -> all_entries
entries -> local_entries
Add a comment:
# An dictionary of { path: SCM url } to describe the gclient checkout.

http://codereview.chromium.org/672004/diff/1/2#newcode849
gclient.py:849: for solution in solutions:
Do you think the code called inside this loop should be refactored into a
function? I'm starting to think it should.

Can you refactor so it uses the same code path than RunOnDeps? I think it'd be
easier to read with smaller functions (or even objects)

http://codereview.chromium.org/672004/diff/1/2#newcode873
gclient.py:873: deps = self._ParseAllDeps(entries, entries_deps_content)
It's confusing since entries_deps_content doesn't need to be a dict anymore.

http://codereview.chromium.org/672004/diff/1/2#newcode1240
gclient.py:1240: 
nit: remove extra whitespace

[jneddo1, 2010-03-09 17:37:54]

Are you suggesting that PrintRevInfo get folded into RunOnDeps?  That makes
sense to me - there's a lot of code duplicated in the two functions.

This is more of a process question; would you want that done for the change to
add this feature, or should folding those functions together be a new bug?

[M-A Ruel, 2010-03-09 18:41:35]

On 2010/03/09 17:37:54, jneddo1 wrote:
> Are you suggesting that PrintRevInfo get folded into RunOnDeps?  That makes
> sense to me - there's a lot of code duplicated in the two functions.

This can be done later.

> This is more of a process question; would you want that done for the change to
> add this feature, or should folding those functions together be a new bug?

You can file a bug if you want to track task that but it's not a requirement.

[jneddo1, 2010-03-09 21:01:24]

http://codereview.chromium.org/672004/diff/1/2
File gclient.py (right):

http://codereview.chromium.org/672004/diff/1/2#newcode847
gclient.py:847: solution_names = {}
On 2010/03/09 15:19:23, Marc-Antoine Ruel wrote:
> What about (from your code):
> solutions_names -> all_entries
> entries -> local_entries
> Add a comment:
> # An dictionary of { path: SCM url } to describe the gclient checkout.

Comments added

http://codereview.chromium.org/672004/diff/1/2#newcode849
gclient.py:849: for solution in solutions:
On 2010/03/09 15:19:23, Marc-Antoine Ruel wrote:
> Do you think the code called inside this loop should be refactored into a
> function? I'm starting to think it should.
> 
> Can you refactor so it uses the same code path than RunOnDeps? I think it'd be
> easier to read with smaller functions (or even objects)

Refactoring this is going to be a decent chunk of work.  The code has branched
off enough that its not quite identical here and RunOnDeps.  

Also, RunOnDeps parses the solutions and DEPS differently now.  RunOnDeps parses
all the base solutions first, then the DEPS files.  To generate a new .gclient
file, I needed to parse a solution, then that solution's DEPS file before moving
onto the next solution.

http://codereview.chromium.org/672004/diff/1/2#newcode873
gclient.py:873: deps = self._ParseAllDeps(entries, entries_deps_content)
On 2010/03/09 15:19:23, Marc-Antoine Ruel wrote:
> It's confusing since entries_deps_content doesn't need to be a dict anymore.

Unfortunately, _ParseAllDeps expects entries_deps_content to be a dict.

http://codereview.chromium.org/672004/diff/1/2#newcode1240
gclient.py:1240: 
On 2010/03/09 15:19:23, Marc-Antoine Ruel wrote:
> nit: remove extra whitespace

Done

[M-A Ruel, 2010-03-09 21:22:49]

lgtm with minor style nits.
Can you upload again with the nits fixed and I'll commit.

http://codereview.chromium.org/672004/diff/6001/7001
File gclient.py (right):

http://codereview.chromium.org/672004/diff/6001/7001#newcode912
gclient.py:912: }
align -2

http://codereview.chromium.org/672004/diff/6001/7001#newcode915
gclient.py:915: for x in sorted(entries.keys())]))
align -1

http://codereview.chromium.org/672004/diff/6001/7001#newcode917
gclient.py:917: #Print the snapshot configuration file
# Print
(space)

[jneddo1, 2010-03-09 21:33:39]

http://codereview.chromium.org/672004/diff/6001/7001
File gclient.py (right):

http://codereview.chromium.org/672004/diff/6001/7001#newcode912
gclient.py:912: }
On 2010/03/09 21:22:49, Marc-Antoine Ruel wrote:
> align -2

Done.

http://codereview.chromium.org/672004/diff/6001/7001#newcode915
gclient.py:915: for x in sorted(entries.keys())]))
On 2010/03/09 21:22:49, Marc-Antoine Ruel wrote:
> align -1

Done.

http://codereview.chromium.org/672004/diff/6001/7001#newcode917
gclient.py:917: #Print the snapshot configuration file
On 2010/03/09 21:22:49, Marc-Antoine Ruel wrote:
> # Print
> (space)

Done.

[M-A Ruel, 2010-03-09 21:40:57]

Thanks, committed.