[openstack-dev] [Nova] Updates to Juno blueprint review process

Thierry Carrez thierry at openstack.org
Tue Mar 25 10:25:07 UTC 2014


Russell Bryant wrote:
> On 03/24/2014 11:42 AM, Stefano Maffulli wrote:
>> At this point I'd like to get a fair assessment of storyboard's status
>> and timeline: it's clear that Launchpad blueprints need to be abandoned
>> lightspeed fast and I (and others) have been sold the idea that
>> Storyboard is a *thing* that will happen *soon*. I also was told that
>> spec reviews are an integral part of Storyboard use case scenarios, not
>> just defects.
> 
> Another critical point of clarification ... we are *not* moving out of
> blueprints at all.  We're still using them for tracking, just as before.
>  We are *adding* the use of gerrit for reviewing the design.

Yes, there is a clear misunderstanding here. There never was any kind of
spec review system in Launchpad. Launchpad tracks feature completion,
not design specs. It supported a link and behind that link was a set of
tools (wiki pages, etherpads, google docs) where the spec review would
hopefully happen.

This proposal is *just* about replacing all those "design documents"
with a clear change and using Gerrit to iterate and track approvals on it.

This is not "moving off Launchpad blueprints" nor is it "bypassing
StoryBoard". It's adding a bit of formal process around something that
randomly lived out of our tooling up to now.


About StoryBoard progress now:

People got very excited with the Django proof-of-concept because it
looked like it was almost ready to replace Launchpad. But I always made
clear this was just a proof-of-concept and it would take a *lot* of time
and effort to make it real. And my usual amount of free time would
probably not allow me to help that much.

It also took us more time than we expected to set up a basic team to
work on it (many thanks to HP and Mirantis for jumping on it with
significant resources), to align that team on clear goals, to rewrite
the main data server as an OpenStack-style API server, to write from
scratch a JavaScript webclient and get it properly tested at the gate, etc.

We now have the base infrastructure in place, we continuously deploy,
and the minimal viable product is almost completed. We expect the
infrastructure team to start dogfooding it in Juno and hopefully we'll
iterate faster next cycle... to make it a viable Launchpad alternative
for adventurous projects in the K cycle.

So it's not vaporware, it exists for real. But there is still a lot of
work needed on it to be generally usable, so it shouldn't be used as an
argument to stall everything else.

-- 
Thierry Carrez (ttx)



More information about the OpenStack-dev mailing list