[MERGE] Fix bug #235407, when the same revision is merged twice

[Robert Collins]

On Mon, 2008-06-02 at 13:46 +1000, Martin Pool wrote:
> On Mon, Jun 2, 2008 at 1:41 PM, Robert Collins
> <robertc at robertcollins.net> wrote:
> > Robert Collins has voted reject.
> > Status is now: Vetoed
> > Comment:
> > merge is definitely doing the wrong thing and should not be any harder to
> > fix: there were tests at one point that add_parent_ids would normalise any
> > new parents so only the heads were listed.
> 
> I believe it does have that check but --force bypasses it.  (I haven't
> checked the code.)  If that's true I can't see any really good reason
> to remove it and would support removing it.  If you can get there
> without --force I'd agree that's a bug.

--force does not bypass the low level tree parent setting code which is
where I believe the bug lies.

> If we want to enforce it in another place you could make commit check
> it, though I think at the moment it avoids scanning history and it
> would be a shame to enforce it.
>
> At any rate: having got a tree into this state, it is not very nice to
> just leave the user with an error.

revert should get the tree out of the state, and then merging with a
fixed bzr would do the right thing.

-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20080602/eff7e2fc/attachment.pgp