On Tue, Oct 6, 2009 at 2:25 PM, Andrew Deason <[email protected]>wrote:

> On Tue, 6 Oct 2009 11:41:35 -0400
> Steven Jenkins <[email protected]> wrote:
>
> > > I don't mean you, particularly. Until we have a framework better
> > > than the one contributed 7 years ago for testing, it's all in flux,
> > > and while I could elaborate suggestions for implementing a new one,
> > > unless someone's actually promising to do it, there are more
> > > pressing and useful things I could do *now* with the time.
> >
> > /me nods.  I don't know where a good, solid win is there either, but
> > if someone *does* know where we could make some solid improvements
> > (i.e., stuff we could get done in the short term, not pie-in-the-sky
> > stuff), I'd like to hear it.
>
> Well, you can still write your own tests, completely ignoring the
> existing framework. I have a few volume-operation-related ones I've been
> using recently. Getting them into a robust state takes time I haven't
> spent yet, though; I don't consider them at the level of quality for
> inclusion in the tree at this time.
>
>
Sure, even sans-framework having tests is valuable, but until we have a
framework, taking them now means more work later when we get something to
fit them into.

Reply via email to