Open Stack

> >
> >
> > What I mean is that we run too into a situation where we have a large
> > backlog of CI jobs since we have to many changes and jobs in flight.
> >
> > So, I'm asking whether there is a good way to not duplicating all jobs
> > to run on all three interpreters. Do we really need testing of all three
> > versions? Or is testing with a subset a manageable risk?
> >
> 
> Fair enough. I'm probably not the right person to answer so perhaps someone
> else can chime in. One thing worth pointing out is that it seems the jump
> from 3.5 to 3.6 wasn't nearly as painful as the jump from 3.6 to 3.7, at
> least in my experience.
> 
> Corey
> 

I share Andreas's concerns. I would rather see us testing 3.5 and 3.7 versus
3.5, 3.6, and 3.7. I would expect anything that passes on 3.7 to be fairly safe
when it comes to 3.6 runtimes.

Maybe a periodic job that exercies 3.6?

Sean