Just running 'bzr selftest blackbox' fails because of Deprecation Warning

[John Arbash Meinel]

Martin Pool wrote:
> On 15 Jun 2006, John Arbash Meinel <john at arbash-meinel.com> wrote:
>> I'm not sure what code path turns off the DeprecationWarning, but if you
>> just run
>> bzr selftest -v blackbox
>>
>> It doesn't disable the warning about all_revision_ids being deprecated.
>> And the merge test checks both stdout and stderr are valid.
>> This is easiest to check with:
>> bzr selftest -v blackbox.*find_merge_base
> 
> I'm not sure why that wasn't seen when running the whole test suite.  In
> any case I have undeprecated that method; refactoring things to depend
> on it less is probably good but deprecating it in the mainline doesn't
> acheive very much right now.

I think the issue is that we only get the deprecation warning one time.
(Or at least a limited number of times). So if you run the whitebox
tests, they don't check stderr for anything, so they don't see the
deprecation warning. But the blackbox tests do check the error output of
bzr, which suddenly has deprecation warnings.

I wish we could deprecate it, but we have a lot of paths that use it
(bzr diff -rancestor:../bzr/bzr.dev).

And as I mentioned before, it sort of conflicts with our desire to fill
in ghosts. Since you might not realize that you can fill in ghosts
unless you combine both histories. (In the limit each side has every
other ghost, so finding one fills in the parents of the other, back and
forth).

John
=:->


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060621/b96e6dea/attachment.pgp