Open Stack

On 02/16/2017 12:20 PM, Ondrej Novy wrote:
> Hi,
> 
> 2017-02-16 0:45 GMT+01:00 Thomas Goirand <zigo at debian.org
> <mailto:zigo at debian.org>>:
> 
>     Yes, you've done some work, and many thanks for it, it has been very
>     useful. However the reality is: since I stopped after the Newton
>     release, absolutely no work has been done to package Ocata in Debian. At
>     this point in the OpenStack dev cycle, normally it should have been
>     fully uploaded *AND* tested against tempest.
> 
> 
> yep, because there are no branches for it. And because I don't know how
> to create them (in infra which I hate for deb packaging) i can't
> continue my work on Ocata.

1/ Let give you the technical answer

Creating a new debian/ocata branch is just one click away. For example,
if you want to do a debian/ocata branch for Nova, just go here:

https://review.openstack.org/#/admin/projects/openstack/deb-nova,branches

type the name of the branch, the sha or simply debian/newton as a ref,
and you're done. That's really super easy, and you even have the ACLs to
do it!

Though the issue here is that you should first get an Ocata repository
created with reprepro. This, only the OpenStack infra team can do. I
investigated quickly, and it seems it should be a "reprepro pull"
command. You'd have to check for it first though, and make sure all
packages gets imported from the Newton pool.

Once that is done, the first package to fix is openstack-pkg-tools, so
that the debian/ocata branch builds and push in the ocata repository.
That's trivial to do, IMO. Just maybe, the pickup job from the infra
team (ie: when a patch is merged) may need to be fixed to push to the
correct place as well.

In any case, the infra team should be able to help, they've been really
helpful.

2/ The social answer

I really wish somebody takes over the work, if I there's no company is
willing to sponsor mine.

But the fact that none (including you) attempted to get in touch with me
to do it over the course of many months, up to the very end of this
Ocata release isn't a good sign. I still don't believe it will happen. I
hope I'll be proven wrong, but I also don't believe it can realistically
happen with just a few volunteers not working on it full time: someone
must get enough commitment to have a global view of what work remains to
be done, do the tests, write patches. That's not something you just do a
few hours on your week-ends. If someone pretends something else, then
this may imply a drastic drop in packaging quality as well, which I'm
*not* willing to put my name on. Otherwise, I would probably have continued.

Also, the approach to give-up the packaging of "non-important" project
(ie: not-core, like the ones in the list I gave...) in the hope it will
save so much time is IMO simply wrong. What eats all of the package
maintenance time is *not* final projects (trove, watcher, etc.) but all
the OpenStack dependencies. Maybe 90% of my time have been spent on
packaging oslo libraries, clients, and 3rd parties, and making sure they
integrate well with the rest of the distro (python 3 compat, Django
1.10, latest SQLA, etc.). So giving-up on that last extra mile of work
to put the cherry on top of the cake isn't the good approach.

Hoping the above helps, though I'm really not sure it does,
Cheers,

Thomas Goirand (zigo)