Re: [Rdo-list] python-*client packaging
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