vc-bzr bogosities

vc-bzr bogosities

[Miles Bader]

I checked out a new emacs and am getting a totally bogus error from
vc-bzr during startup (when desktop.el is loading various files):

   Warnings in `bzr' output: Unable to load plugin 'conflicts' from '/usr/lib/python2.4/site-packages/bzrlib/plugins'
   bzr: ERROR: Not a branch: /home/soft1/miles/some-dir/some-dir-20070711/

   Desktop: Can't load buffer some-dir-20070711: Running bzr revno some-dir-20070711...FAILED (status 3)

Note only isn't that a bzr dir, but I don't even _use_ bzr (though I
have a .bar config file from a long time ago, which only has my email
address in it).

Subsequently, when trying to report this bug, vc-bzr starting spazzing
out and giving errors on all sorts of apparently random files:

   Parsing /home/soft1/miles/.mailrc... done
   Warnings in `bzr' output: Unable to load plugin 'conflicts' from '/usr/lib/python2.4/site-packages/bzrlib/plugins'
   bzr: ERROR: Not a branch: /home/soft1/miles/me/sigs/misc/

   vc-do-command: Running bzr revno misc...FAILED (status 3)

["...sigs/misc/" is my directory of random mail signatures!]

Thanks,

-miles

-- 
A zen-buddhist walked into a pizza shop and
said, "Make me one with everything."

Re: vc-bzr bogosities

[Stefan Monnier]

> Note only isn't that a bzr dir, but I don't even _use_ bzr (though I
> have a .bar config file from a long time ago, which only has my email
> address in it).

I guess the above ".bar" was meant to be ".bzr" and is really ~/.bzr, right?
If so, that explains it: VC walks up the directory tree until it finds
a ".bzr" subdirectory and considers it to be the root of a BZR tree.

We need to make this determination more selective.  I don't know BZR nearly
well enough to know what should be done.  Would checking for
.bzr/branch-format be a good check (i.e. would it always succeed in a real
BZR workspace and would it normally fail in the ~/.bzr config dir)?


        Stefan

Re: vc-bzr bogosities

[Riccardo Murri]

On 7/24/07, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
> > Note only isn't that a bzr dir, but I don't even _use_ bzr (though I
> > have a .bar config file from a long time ago, which only has my email
> > address in it).
>
> I guess the above ".bar" was meant to be ".bzr" and is really ~/.bzr, right?
> If so, that explains it: VC walks up the directory tree until it finds
> a ".bzr" subdirectory and considers it to be the root of a BZR tree.
>
> We need to make this determination more selective.  I don't know BZR nearly
> well enough to know what should be done.  Would checking for
> .bzr/branch-format be a good check (i.e. would it always succeed in a real
> BZR workspace and would it normally fail in the ~/.bzr config dir)?
>

That would probably be a workaround, although the only authoritative
check would be to run `cd file-name-directory && bzr root` (but that
slows vc-bzr down a lot).

Riccardo

Re: vc-bzr bogosities

[Stefan Monnier]

>> > Note only isn't that a bzr dir, but I don't even _use_ bzr (though I
>> > have a .bar config file from a long time ago, which only has my email
>> > address in it).
>> 
>> I guess the above ".bar" was meant to be ".bzr" and is really ~/.bzr, right?
>> If so, that explains it: VC walks up the directory tree until it finds
>> a ".bzr" subdirectory and considers it to be the root of a BZR tree.
>> 
>> We need to make this determination more selective.  I don't know BZR nearly
>> well enough to know what should be done.  Would checking for
>> .bzr/branch-format be a good check (i.e. would it always succeed in a real
>> BZR workspace and would it normally fail in the ~/.bzr config dir)?
>> 

> That would probably be a workaround, although the only authoritative
> check would be to run `cd file-name-directory && bzr root` (but that
> slows vc-bzr down a lot).

Not just that, but the original problem is that running `bzr' signals errors
so my proposition is trying to reduce the cases where we need to run `bzr'.


        Stefan

Re: vc-bzr bogosities

[Dan Nicolaescu]

Miles Bader <miles.bader@necel.com> writes:

  > I checked out a new emacs and am getting a totally bogus error from
  > vc-bzr during startup (when desktop.el is loading various files):

Here's another one: 

mkdir /tmp/test
cd /tmp/test
bzr init

Now in emacs 

M-: (vc-bzr-registered "/tmp/test/foo") RET
up-to-date

(Obviously /tmp/test/foo does not exist...)