Comment 7 for bug 476293

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

Ian Clatworthy wrote:
> John A Meinel wrote:
>> Vincent Ladeuil wrote:
>>> I've implemented a simple plugin (lp:~vila/+junk/bzr-graph) to answer the original question:
>>> is a revision an ancestor of another one.
>>> Contrary to my comments, 'bzr log -rX..Y' now requires X to be older than Y.
>>> The case exposed here is a genuine bug as 'bzr log -n0 --show-ids -r tag:mysql-5.1.39..tag:clone-5.2'
>>> should fail instead of displaying irrelevant revisions.
>>
>> I think Ian mentioned that it was a design decision to start logging
>> what we could, when we couldn't (yet) determine whether the arguments
>> were correct. So rather than searching all of history to say that they
>> args are invalid, and *then* displaying revs, start displaying revs
>> until we can determine whether it is valid or not.
>
> That was my thinking at the time. In hindsight, I think it's fine to
> validate the lower bound before displaying revisions, given that
> explicitly providing a lower bound is relatively rare.
>
> Ian C.
>

Well, I do it by default... (bzr alias log=log --short -r -10..-1)

I suppose giving a lower bound such that it is sufficiently old to cause
perf problems is rare...

John
=:->

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

iEYEARECAAYFAktXga0ACgkQJdeBCYSNAANzlwCdF/yhfSzALqB/0qm8Kt+yUuXV
ffoAoNcf1osuxmnbCdm38yVbKqObK1S/
=IIMe
-----END PGP SIGNATURE-----