Re: [gentoo-dev] Request for testing

[Karl Trygve Kalleberg]

On Mon, 19 Aug 2002, Wilbert van Bakel wrote:

> What are we testing? What are the guide lines? Which quality level do we
> want to reach? Which USE flags do you want to include in the test? Which
> compiler?
> Does Gentoo approve an update when software is emerging without
> problems, or when software is known to be productive without problems?
> And... How do we report bugs, which information is actually useful? And
> do we report succesful installations too?

These are all crucial and interesting questions if the Gentoo Project is
to develop any kind of formalised test procedure. We don't have answers to
them at the current time, and that's also one of the reasons why we are
(and rightly should be, IMHO) considered a "hobbyist" distro.

We're slowly phasing in the focus on quality in addition to our up-to-date
focus. The two are definitely competing features, and currently
"up-to-date" is leading the race pretty well.

As part of that focus shift, we'll have to decide where our sweet spot
between bleeding edge and rock-steady, super-tested lies. Chances are,
we'll always be way off against bleeding edge, but we can implement
heaploads of simple measures to ensure that all packages in our Tree work
more of the time. 


Currently, the enthusiasm is about getting 1.4 out the door (the most
exact time of release I can give you is "around tea time"). Bart and his
merry men are already cutting their teeth making that a less painful
affair for the upcoming 2.0 release (probably "around supper time"). 

I expect that once the minor nuisance of pushing 1.4 out the door is dealt
with, we'll pool our experiences and start formulating some form of 
testing policies.


So in summary: 
* Yes, we want testing to be more systematic.
* Yes, we know testing is non-trivial.
* Yes, it runs with Novell Netware.


Kind regards,

Karl T