"bzr pull -r 10000" doesn't pull entirety of 9998-rev branch

[Samuel Bronson]

naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2% bzr pull -r 10000 && bzr pull -r 11000 && bzr pull
Using saved parent location: svn://scm.gforge.inria.fr/svn/coq/branches/v8.2
bzr: ERROR: Requested revision: u'10000' does not exist in branch: SvnBranch('svn://scm.gforge.inria.fr/svn/coq/branches/v8.2')
naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2% bzr pull
Using saved parent location: svn://scm.gforge.inria.fr/svn/coq/branches/v8.2
[pulls 998 revisions that it didn't have before ...]
Now on revision 9998 (svn revno: 12299).
naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2%

It would have been much more intuitive if the first pull had pulled up to revision 9998 and simply *pointed out* that there wasn't actually a revision 10000, rather than refusing to pull anything.

[John A Meinel (jameinel)]

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

Samuel Bronson wrote:
> Public bug reported:
>
> naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2% bzr pull -r 10000 && bzr pull -r 11000 && bzr pull
> Using saved parent location: svn://scm.gforge.inria.fr/svn/coq/branches/v8.2
> bzr: ERROR: Requested revision: u'10000' does not exist in branch: SvnBranch('svn://scm.gforge.inria.fr/svn/coq/branches/v8.2')
> naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2% bzr pull
> Using saved parent location: svn://scm.gforge.inria.fr/svn/coq/branches/v8.2
> [pulls 998 revisions that it didn't have before ...]
> Now on revision 9998 (svn revno: 12299).
> naesten@hydrogen:~/hacking/haskell/coq-bzr-2a/branches/v8.2%
>
> It would have been much more intuitive if the first pull had pulled up
> to revision 9998 and simply *pointed out* that there wasn't actually a
> revision 10000, rather than refusing to pull anything.
>
> ** Affects: bzr
> Importance: Undecided
> Status: New
>

You could always do "bzr pull -r -1" and always pull to the last revision.

I would probably say that the error message could say:

revno 10000 does not exist, the last revision is: YYYY

which might work as well.

I don't think we want to pull to a random revision and have the user not
know where they are really at.

Giving a wrong revision is usually an indication that you either aren't
fetching from the branch you think you are, or you somehow have a wrong
indicator.

John
=:->

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

iEYEARECAAYFAkqiqywACgkQJdeBCYSNAANtDwCghBevrgDfgNu6gDWA4+oiX7vl
KzIAn2XBxofadI12IrIjrP7rMpL/alNH
=Y9mg
-----END PGP SIGNATURE-----

[Samuel Bronson (naesten)]

Ah, well, the reason I was doing that in this case was that I was pulling in blocks of 1000 or so, so that I would not end up with nothing if my computer happened to reboot in mid-pull. This might sound silly, but it had already happened once for that branch, leaving me with no revisions in the repository, so I wasn't taking any chances...

[Vincent Ladeuil (vila)]

The wish is that the error message could say:

revno 10000 does not exist, the last revision is: YYYY

which might work as well.