Adding Tony Breeds candidacy for Requirements

Change-Id: I8a029205c517c1d3dafde5435facf7c5be2ad97f
This commit is contained in:
Tony Breeds 2017-01-29 14:10:24 +11:00
parent ad5f7cc837
commit 5e306a4732
1 changed files with 54 additions and 0 deletions

View File

@ -0,0 +1,54 @@
I'd like to continue to serve as the OpenStack Requirements PTL.
The focus for the Ocata cycle was on establishing tools to maintain
a co-installable set of projects whilst given each project more control
in their own repos to set minimums and banned versions. That work was
predicated on the statement[1]:
"nearly everyone is now using constraints"
It turns out that when that state was investigated it was untrue. It seems
about 50% of projects listed projects.txt[2]. So the focus switched to
enabling tools and projects to use constraints[3].
So we've set the groundwork to attack the 2 biggest issues with requirements
for OpenStack.
1. Divergent requirements
Allowing projects to be co-installable but not requiring projects to
maintain test identical requirements.
2. Testing lower-bounds
In my Ocata platform I called out the following items:
1. Improving communication, often decisions are made in the requirements team
that affect many projects, I have a commitment to bringing more experts in
for strategic reviews/discussions
2. Understand how we can work with/enhance pip tooling to create a more
satisfactory requirements/constraints experience for OpenStack
3. Work closely with the release managers as there is still a lot of common
issues there. In that a release of $project will trigger processes in the
requirements team.
4. Getting openstack_requirements *code* to the point it can be installed as a
library. We've seen issues in the past where stable branches need largish
backports to work correctly. Really the *code* and *data* should be treated
independently
5. Testing lower bounds of our supported requirements [lower-constraints.txt]
By way of an update/clarification of what we achieved and what's left to do.
1. We've done some of this but what's really missing is ways to rapidly
identify whom the outside SMEs should be and reach out.
2. We've put in place a number of band aids here but the pip enhancements
are "big a scary"[4].
3. Ongoing. Frankly I dropped the ball here but as always the Release
Managers team have been fantastic here.
4. Done. We're yet to leverage this fully.
5. See opening pre-amble :)
I'd appreciate another cycle to progress these items, if y'all will have me.
Yours Tony.
[1] /me during the Barcelona summit (more than once) ;/
[2] http://git.openstack.org/cgit/openstack/requirements/tree/projects.txt
[3] We're now in excess of 85%
[4] My assessment not that of the pip team.