Open Stack

On Wed, 25 May 2016, Sean Dague wrote:

> I still would rather not put gabbi into the compute API testing this
> cycle. Instead learn from the placement side, let people see good
> patterns there, and not confuse contributors with multiple ways to test
> things in the compute API. Because that requires a lot of digging out
> from later (example: mox & mock).

To be clear, I wasn't saying "let's do this immediately" or even
"let's do this this cycle". What I'm trying to do is two things. One
is to lay, slowly, some groundwork on which we can build up an
understanding of two things:

* what gabbi can do
* which of those things might be useful for nova

That's a conversation that can carry on pretty slowly and doesn't
have to take away from anything else. But as I've been noticing a
lot lately, if we try to go into changes without having some
agreement on the words we're using, we're not going to get anywhere,
so you know, let's have a chilled chat about this stuff and see
where it takes us. That's an important part of the process and the
medium of email is a reasonable place for that process (inclusive,
asynchronous, addressable).

The other is to grant people who do have the wherewithal to improve
their stuff (be that stuff nova or something else) with gabbi some greater
visibility into gabbi's existence and prowess. Knowing is half the
battle, etc.

> And we still have this whole api-ref site which is only 50% verified
> (and we still need to address a number of microversion issues) -
> http://burndown.dague.org/. We said at the beginning of the cycle
> api-ref and policy in code were our 2 API priorities. Until those are
> well in the bag I don't want to take the energy and care to make sure we
> do a pivot on test strategy to something completely new in a way that is
> easy for everyone to contribute to and review.

a) I promise to be good boy and get involved there. I keep meaning
    to and a variety of other things keep coming up (including simply
    the need to be in a different zone to clear out the crazy) and I
    feel lame about it.

b) I'd like to disabuse you of this notion that there is a pivot
    involved here or being suggested here. I prefer to think of it as
    an augmentation.

    However, even if it is a pivot: So what? Sometimes we need to
    make changes. Sometimes because it is necessary and we need new
    functionality. Sometimes simply because changing things up a bit
    provides a _much_ needed shift in perspective.

-- 
Chris Dent               (╯°□°)╯︵┻━┻            http://anticdent.org/
freenode: cdent                                         tw: @anticdent