Open Stack

On Fri, Feb 15, 2019 at 2:42 AM Artem Goncharov
<artem.goncharov at gmail.com> wrote:
> Dean, are you ok in receiving changes for switch to SDK before we get R1.0?

I suppose we should just go ahead and do that.  We've been burnt twice
since we started doing the Network commands by changes, there was a
lot of compatibility stuff on both sides for the last one, I don't
want either project to have to do that again.  Monty told me a while
back that he didn't expect any more compat-impacting changes, if that
is still true I'd say we should start...

> Let us really just focus on few services as a target and then hopefully achieve more. What do you think?

> My suggestion would be to focus on:
> - novaclient
> - glanceclient
> - swiftclient

I had been planning to do glance first for a number of reasons, one
being the number of unique dependencies glanceclient bring in to OSC.

Swift is a special case, we don't use swiftclient at all, I used what
I had originally proposed to the SDK (a really long time ago now) for
a low-level API and copied the useful functions directly from swift
(this was even before swiftclient was a thing).  The core of what is
in OSC for swift is swift code, way out of date now, which is one
reason so much of that API is not implemented.

Both of those have notiecable returns to do first although it could
also be argued that due to the above history of swift support in OSC
not a lot of users are relying on that.

For this to make sense as a community goal, I want to support projects
that want to do similar things themselves too, although much of that
is in plugins.  Being around for spiritual support and getting the SDK
updated would be the need from my point of view here.

dt

-- 

Dean Troyer
dtroyer at gmail.com