Re: [h5md-user] The first H5MD modules: "particles" and "observables"

[Pierre de Buyl]

On Mon, Oct 07, 2013 at 12:08:37PM +0200, Felix Höfling wrote:
> Am 07.10.2013, 11:52 Uhr, schrieb Konrad Hinsen
> <address@hidden>:
> 
> >Let me take my pragmatist's role again and ask the question: what
> >difference does it make whether "particles" and "observables" are
> >part of the core definition, or optional modules?
> >
> >The only difference I see is that in the first case, the two groups
> >must exist even if they are empty, whereas in the second case one or
> >the other may be missing completely.
> >
> >If that's indeed the only practical difference, I suggest throwing
> >a coin to settle the issue and move on to more important subjects.
> >
> Actually, both groups are optional in the current draft. There is _no_
> difference whether we call them modules or not. It is just a matter of how
> the H5MD spec is presented. Calling them modules would be clearer
> conceptually (at least to me) and it would give life to the concept of
> modules. But others might have objections.
> 
> Regarding "more important subjects": I agree that we urgently need a
> priorisation of the various topics discussed on the mailing list (some of
> the threads have apparently died albeit the issues have not been solved).
> A kind of bug tracker would be of great value. Maybe Github could to the
> job?

I am happy to announce that the problem is settled and that modules are not used
for "particles" and "observables". It will simplify everyone's life when using
H5MD. Let me remind that H5MD is for molecular data at first and that we are not
preparing a "mega super generic file format for all kind of scientific data".
That format is HDF5. Our format is aimed, at first, at molecular data. It so
happens that it is a rather generic format but its first use is to store the
kind of data that goes into "particles" and "observables". Further genericity
could come in the future.

Cheers,

Pierre