Open Stack

On 8/31/2015 4:30 PM, Jeremy Stanley wrote:
> On 2015-08-31 16:23:14 -0500 (-0500), Matt Riedemann wrote:
>> I've released 0.10.0:
>>
>> https://pypi.python.org/pypi/sqlalchemy-migrate/0.10.0
>
> With an intermediate 0.9.9 you could have avoided leaving things
> broken while requirements gets updated. Consider that a "revert" of
> tagging.
>
>> I have a patch up to g-r on stable/kilo to block 0.9.8 and cap at <0.10.0:
>>
>> https://review.openstack.org/#/c/219027/
>
> Would have been nice to get that merged first with a "revert" of the
> 0.9.8 tag (per above).
>
>> I don't think we need to revert that change.  sqlalchemy-migrate doesn't
>> have a stable/kilo branch, but if/when we need one, I'd create it from the
>> 0.9.7 tag.
>
> Sure, and tag it... what? There's nothing higher than 0.9.7 which
> wouldn't require yet another change to the requirements cap in kilo
> (and subsequent propagation).
>
>> I also pushed a change to the infra-manual docs to remind me from being dumb
>> in the future:
>>
>> https://review.openstack.org/#/c/219030/
>
> Thanks, +2!
>

Good points, we should have talked in IRC since I didn't see this until 
after doing the 0.10.0 tag, otherwise tagging 0.9.8 as 0.10.0 and 0.9.7 
and 0.9.9 would have fixed things w/o the g-r sync.  :)  mtreinish made 
the same point when I asked him to review the g-r change in kilo.

I guess I was only thinking of moving forward with the 0.10.0 for 
liberty and then blacklist 0.9.8 - act as if it never happened.  That 
still requires the syncs to happen, which sucks, but lesson learned 
(hopefully).

The next release on a stable/kilo branch for sqla-migrate would be 
0.9.9.  It will be a bit weird, but I think that's what it'd have to be.

-- 

Thanks,

Matt Riedemann