[Rdo-list] python-*client packaging

[Steven Hardy]

On Mon, Jun 02, 2014 at 10:08:41AM -0700, Dan Smith wrote:
> > It's true that major regressions normally get attention pretty quickly, but
> > more often what happens is the consuming projects, such as Heat, fix their
> > master to cope with the new client behavior to un-break their gate and move
> > on.  Again, this implies that by picking a random git version of a client,
> > you run the risk that various projects will just break in exciting ways.
> 
> Yeah, it's too bad that you do that. If we don't have any indication
> that something broke, it's hard to know to fix it :)

I never said we don't report the issues, I even linked a novaclient bug
earlier in the thread, we always report regressions when we find them :\

Sometimes the issues are bugs in heat, most often coupling between our unit
tests and the client, similar issues are sometimes encountered by Horizon,
e.g bug #1279907.  Arguably we need a better way of stubbing client
interfaces to avoid this problem.

My point remains that it's preferable to work with upstream to trigger
a new release, but if folks want to run bleeding edge clients then fair
enough.

Steve