[PATCH] 306394 fix (warning, not error, on non-existent file)

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Karl Fogel wrote:
> Aaron Bentley <aaron at aaronbentley.com> writes:
>> Did you ever commit to bzr.dev-trunk?  It seems like you committed a
>> revision called "kfogel at red-bean.com-20090108210552-4cv8mvc6sv0gdbn1".
> 
> Yes, I did.  That was (meant to be) a merge from upstream mainline:

...

>> $ echo "public_branch = http://bazaar-vcs.org/bzr/bzr.dev" >>
>>   bzr.dev-trunk/.bzr/branch/config
> 
> Done.  (I admit I'm not quite clear on why this is necessary; Googling
> and trawling bazaar-vcs.org hasn't totally enlightened me yet.)
> 

This sets the "submit_branch:" in the bundle to something other than
your local directory. Bundle buggy can use that to pull extra revisions
that did not end up in the bundle.

It still would have failed in this case, because the public location
isn't actually an exact match for your local branch (because you merged
rather than pulled.)

>> If I'm right that you committed into your local mirror, that's not sane.
>>  (But you can fix it with "bzr pull --overwrite -d bzr.dev-trunk")
> 
> Done, thanks (and 'bzr log' confirms that the offending revision is
> gone).
> 
> -Karl
> 
> 

Personally, I find that a checkout is good for my "bzr.dev" branch.
Because that ensures that I can't commit to it. You can get there with:

cd bzr.dev-trunk
bzr bind http://bazaar-vcs.org/bzr/bzr.dev

If you ever try to 'bzr commit' in that branch, it will tell you that it
cannot lock an HTTP transport.

At that point, you use "bzr update" to update your copy, which also
(IMO) feels like a 'update my mirror' rather than merge or pull (though
pull is also update-my-mirror, just for non-checkouts.)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkltEScACgkQJdeBCYSNAANf6QCfeWGxCMm1e14UCPf1G4Nlo1G3
ug4AnAouLOSPrwLHSaSvvy+ihP1t7/ad
=PHg5
-----END PGP SIGNATURE-----