[PATCH 5 of 5 VERSION2] subrepo: removing (and restoring) git subrepo state

[Augie Fackler]

On Nov 14, 2010, at 8:56 PM, Eric Eisner wrote:

>> 
>>> +    def remove(self):
>>> +        if self.dirty():
>>> +            self._ui.warn(_('not removing repo %s because '
>>> +                            'it has changes.\n') % self._path)
>>> +            return
>>> +        # we can't fully delete the repository as it may contain
>>> +        # local-only history
>>> +        self._ui.note(_('removing subrepo %s\n') % self._path)
>>> +        self._gitcommand(['config', 'core.bare', 'true'])
>> 
>> Why mark a repo as bare? I suppose that emulates hg co null, but that seems
>> like a strange thing to do. I'm not sure if we can do better though, so I
>> guess it's just something we have to live with.
>> 
> 
> Git has no direct equivalent to checking out null, but marking the repo bare
> prevents any operations that affect the working dir.
> 
> 
>> 
>>> +        for f in os.listdir(self._path):
>>> +            if f == '.git':
>>> +                continue
>>> +            path = os.path.join(self._path, f)
>>> +            if os.path.isdir(path) and not os.path.islink(path):
>>> +                shutil.rmtree(path)
>>> +            else:
>>> +                os.remove(path)
>> 
>> What's this code doing? Just trying to understand.
> 
> 
> Simply marking the repo bare doesn't actually touch the working dir, so I
> explicitly remove all files that are not the .git directory.

Ah, that's surprising. I'd like a comment in this code to that effect so that future self knows what's going on.

> 
> 
> [forgot the reply-all, sorry Augie]