[MERGE] Shelf 5 / 5 for reals

[Aaron Bentley]

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

John Arbash Meinel wrote:
>>> Do we want to be raising SystemExit, rather than a more bzr-specific
>>> exit exception?
>> Well, it's very direct, and this is commmand-level functionality, so it
>> felt okay to me.  But I've switched that to raise UserAbort.

> Most commands that decide they need to stop raise BzrCommandError. I
> realize this isn't an error as much as "user-requested we stop".

Exactly.  It's not an error at all, and shouldn't produce an error
return code at the commandline.  I would have to catch it as I am
presently doing with UserAbort, so it would have to be a subclass of
BzrCommandError, not BzrCommandError itself.

> I know the existing shelf aborts with a strange:
> 
> bzr: ERROR: nothing to shelve
> 
> When there is
> a) No changes
> and
> b) User hits cancel
> 
> So I don't suggest that we mimic that behavior exactly.

Okay, so what do you suggest we do?

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkkQVMUACgkQ0F+nu1YWqI21qgCffC4eWqsJNzniu1+e1pmYJ66j
KA0An24QLFz3T+nVndKvGgEOtOLJdvKB
=4k/G
-----END PGP SIGNATURE-----