Discussion on cp2k-release-supported-features-wishlist

Axel akoh... at gmail.com
Wed May 21 17:38:35 UTC 2008

On May 21, 2:57 am, ilya <ily... at gmail.com> wrote:
> Hi!
> I think that you should support unique features of that program
> primarily. To my mind, the more unique the feature the more important
> it should be.
> So the list is something like metadynamics, GPW, GAPW, qm/mm, CSVR,
> etc + some algorithms to perform basic operations such as
> optimizations.

in principle you would be correct, but where do we
find a "superhero" that can document _and_ test all of
those features all at the same time? it is already a
remarkable achievement that cp2k has reached its current
level of reliability despite the rather high speed of
the ongoing development. however, seldomly used components
that are not fully checked via regtests and odd combination
of features (the incredible flexibility of cp2k is a big
problem, hence the self-limitation) tend to get broken
unintentionally every once in a while. for a release we
need to be able to say "we tested these features well and
they work ok so far". for people that want more there is
always the cvs version and then they have to cope with
changing interfaces, occasionally changing behavior and
fresh bugs.

_helpful_ documentation is the part that is currently missing
the most (just have a look at the questions here), so my
next step is going to write a documentation skeleton and
then ask people here to help adding to it. despite popular
belief, beginners are the most qualified to provide this
kind of documentation, as developers tend to overlook
seemingly trivial issues and use "slang" and in any case
should spend their time and what they do best: improve the code.


> Ilya.
> On May 19, 1:15 am, Axel <akoh... at gmail.com> wrote:
> > Hi!
> > here's a first sketch of what would be a list of goals for the first
> > release.
> > please adjust and/or put comments here.
> > thanks,
> >      axel.

More information about the CP2K-user mailing list