On 26/08/15 12:22 -0400, Nikhil Komawar wrote:
We considered that option and have finally agreed on what Alex suggests
at the rel-mgrs office as this is the least painful path for most.

It would have been nice to know it had already been discussed across
some glance members and other folks. I'm sure you all considered other
options too so I'm good.

It'd be nice to have a link to the discussion too.

Thanks,
Flavio



On 8/26/15 12:04 PM, Flavio Percoco wrote:
On 24/08/15 18:37 +0300, Alexander Tivelkov wrote:
Hi folks,

In the upcoming L release Murano is going to use the "Glance Artifact
Repository" feature implemented as part of EXPERIMENTAL Glance V3 API.

The server-side support of this feature is already merged in glance's
master
branch, while the client code is not: it was agreed that the v3's
client will
stay in a dedicated feature branch (feature/artifacts) and will not
be released
until the API is stable and final (a major API refactoring based on the
feedback from API WG is on the way and is likely to happen in M). So,
there
will be no v3-aware releases of python-glanceclient on pypi until
then. The
early adopters of V3 API are encouraged to build the tarballs out of the
feature branch on their own and use them, keeping in mind that the
API is
EXPERIMENTAL so everything may be (and will be) changed.

However, Murano needs some way to consume Glance V3 right now, and  -
as it has
a voting requirements job at the gate - it cannot just put a git branch
reference in its requirements.txt. It needs some kind of a release
which would
be part of global requirements etc.

So, it was decided to temporary copy-paste the experimental part of
glance
client into the murano client (i.e. to copy python-glanceclient/feature/
artifacts/v3 to python-muranoclient/master/artifacts) and release it
as part of
several next releases of python-muranoclient.

When the Glance V3 API is stable, we'll put its client to the master
branch of
python-glanceclient and release it normally, then dropping the
temporary copy
from python-muranoclient.

Until then, all the changes to the experimental client should be done
in the
feature/artifacts branch of glance client and copied (synced) to
murano client.
Similar to oslo.incubator sync procedure, just without a shell script :)

I hope that the need to do this copy-pasting will not last for long
and the v3
API becomes stable soon enough.

I was going to sugest having a separate library just for this code.
You'd work on this library until the API is stable and then you'd
merge it back into glanceclient as soon as the feature is stable
server side.

We could also have a way to load the library code as a plugin for
glanceclient - similar to the way openstack client works - but that
requires a spec for mitaka.

Anyway, the library could be called `python-glanceclient-artifacts` or
something along those lines.

Flavio



__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

--

Thanks,
Nikhil


__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

--
@flaper87
Flavio Percoco

Attachment: pgpd4IbvyuGxn.pgp
Description: PGP signature

__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to