Open Stack

Excerpts from Monty Taylor's message of 2017-06-16 14:15:51 -0500:
> Hey all,
> 
> We have an important balancing act going on right now, and we also have 
> an increasing number of lovely humans playing in our playbox. (hello 
> lovely people!)
> 
> The balancing act is that we're in the final push of getting this bad 
> boy rolled out, and we're having to keep a tight focus on that initial 
> finish line. BUT - we did purposely design Zuul v3 to be friendly for 
> other folks to run and extend, so it's TOTALLY REASONABLE for folks to 
> be interested in working on things that fall into that area.
> 
> We need to both empower those friends, but make it clear that while we 
> support their needs and their enthusiasm, we're making a conscious 
> decision to keep our focus on the core for the next little bit so that 
> we can get it out and be in a good position to spend more time working 
> with folks on new things.
> 
> For anyone who is just starting to pay attention (or as a reminder for 
> the rest of us) the current top-priority list is:
> 
> - Get it Rolled out for OpenStack
> - Get BonnyCI running with no local changes
> - Document use and operation well to empower consumption
> - Get it Rolled out for Ansible
> 
> There are a few other things that aren't strictly required for those but 
> that we've all pretty much agreed are still in-scope for current effort. 
> Supporting them is still lower priority than direct work on the above, 
> but it's totally in scope and should get review bandwidth/engagement.
> 
> - Define Nodepool Provider Plugin Interface
> - Implement Static Node support for Nodepool
> 
> Then there's a TON of stuff on the radar we'd love to be in Zuul or that 
> we know we need to solve or do. It seems to me that we need a list of 
> those things so that people know:
> 
> a) Yes, we're aware
> b) Yes, we also think it's cool and desirable
> c) Sorry, we're purposely not touching it today
> 
> Some of these things may get stories in storyboard - but honestly some 
> of them are just placeholders indicating that we need to have a 
> discussion about them later - or that we'll write a story later when we 
> know something.
> 
> I propose putting a list into the zuul repo in a ROADMAP.rst file. It's 
> an easy place for us to keep track of the efforts that are under way and 
> the ones that we know are a thing and want to do but are purposely not 
> working on yet.
> 
> I don't think this is a replacement for the use of storyboard - but 
> since the majority of this list is "not even designing it yet" stage, I 
> think a quick summary that's easy to reference is in order.

I am not so sure I like this. There's a value in storyboard that
ROADMAP.rst will lose, which is that it is flexible and can tie in to
other projects, which is the main reason Storyboard exists and isn't
just another Bugzilla. It would be a shame to turn our backs on it for
the exact job it was built to do.

Here's an alternative plan...

This has been our work-centric roadmap since October-ish:

https://storyboard.openstack.org/#!/board/41

Specifically the Todo column should be only the things we are definitely
going to do before we put zuulv3 in production in infra (we've discussed
whittling this down a bit actually).

But this is focused on those developing in the immediate. It is entirely
built from the 'zuulv3' tag, and then hand-organized between Todo/In
Progress/Blocked during meetings or normal course of development.
Completed tasks and stories are removed automatically.

So how about we make a new tag, zuul-roadmap. Anything that we need to
have a conversation about and are just forming opinions on can be tagged
with this tag. If and when we start writing specs or trying to design it,
we can add the zuulv3 tag (or whatever tag we're using at the time for
"the stuff we're doing now") and remove it from zuul-roadmap when we
finish. This would be relatively simple, and keep the value of having
our planning happening in one system and gives us one place to answer
"what's going on in zuul?"