Re: Bug with large output on vc commands

[Kai Willadsen <kai willadsen gmail com>]

On 13 December 2012 00:54, louis <louis obsidian com au> wrote:
>
> On 12/12/2012 8:02 AM, "Kai Willadsen" <kai willadsen gmail com> wrote:
>>
>> So looking at history, we already changed from 'check' to 'status' a
>> while ago to make it faster.
>>     https://bugzilla.gnome.org/show_bug.cgi?id=653302
>>
>> The idea is that valid_repo() should tell us whether we're actually
>> operating in something that bzr recognises as a valid repository. Do
>> you think that 'bzr info' is good enough as far as that kind of check
>> goes? Basically, we want to know that further bzr calls on that path
>> will actually do something sensible.
>
> Info is good enough and *much* quicker imo.
> It simply returns parent paths of this branch and so forth.
>
> Other options: root, version-info
>
> If any of these commands succeed, then there is a .bzr directory somewhere
> in our above this folder, status won't be doing any more about checks than
> these commands, check is the only command that does integrity checks.

(The following is basically FYI for interested parties.)

The point of this call is not to establish that there's a .bzr
directory there; we have find_repo_root() for that. The point is that
valid_repo() should tell us whether there is *actually* a repo there,
and whether it's a repo that the VC can handle. For example, can the
VC binary that we're calling actually handle the repo format? Calling
SVN 1.6 on a SVN 1.7 repository will fail, and the valid_repo() call
is there to pick up that kind of case.

Anyway, looking at what info interrogates, I'm pretty happy that it
provides the basic checks we want, so I'm okay to switch over. I've
just pushed the change to head.

cheers,
Kai