Open Stack

On Mon, Jul 14, 2014 at 7:38 AM, Duncan Thomas <duncan.thomas at gmail.com>
wrote:

> On 14 July 2014 07:11, Flavio Percoco <flavio at redhat.com> wrote:
> > I almost fully agree with this last point. The bit I don't agree with is
> > that there are some small refactor changes that aim to change a core
> > piece of the project without any impact on the final user that are
> > spec/blueprint worthy to explaining the motivation, expected results and
> > drawbacks.
> >
> > To put it in another way. Developers are consumers of project's code,
> > therefore the changes affecting the way developers interact with the
> > code are also blueprint worth it, IMHO.
>
> The way I've been playing it on cinder is to ask for a spec if I'm
> reviewing a patch that doesn't have one and I find myself questioning
> the approach rather than the code.
>

Exactly.

Also in Ironic, we're still feeling around when a spec is appropriate
versus when it's just a bug, and I know we've had a few cases where both
were created multiple solutions for the bug were found and I (or others on
the core team) wanted some discussion around and justification for the
proposed solution, the specs process seemed like a better medium for that
discussion than the launchpad bug page.

Of course, discussions like this one help us all to think about and find a
good balance -- not every change requires a spec, but, as Flavio pointed
out, sometimes even just refactoring code has enough impact on developers
that a discussion would be beneficial. (I'm thinking of the nova db object
code as one example...)

Best,
-D
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140714/ad616041/attachment.html>