Re: [h5md-user] H5MD for proteins

[Konrad Hinsen]

Olaf Lenz writes:

 > I had thought that the HDF5 DTD might provide that
 >   http://www.hdfgroup.org/HDF5/XML/dtd-info.html
 > 
 > Unfortunately, it just seems to provide a DTD that describes the
 > possible output of "hdf5dump -x".
 > 
 > Maybe we can use "hdf5dump -x" in conjunction with an XML validator to
 > validate H5MD-files?

That's worth exploring, though I have doubts that the amount of
verification that can be done is worth the effort. XML validation
checks only form, not content, and the output format of h5dump adds
another layer of form that makes it even more difficult to check the
interesting stuff.

The biggest limitation of XML validators is that they check one
element at a time. They cannot check, for example, that the shape of
the "positions" array is compatible with the shape of the "time"
array.


I looked at pyh5md today and noticed that it has a validation function,
which looks like a good start for a H5MD validator.

 > I propose to add semantic information step-by-step via modules. Like
 > this we can maybe avoid to render the format unusable for some
 > applications. That is why I would also vote to keep the base module as
 > simple as possible.

I like that approach a lot.

Konrad.
-- 
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: research AT khinsen DOT fastmail DOT net
http://dirac.cnrs-orleans.fr/~hinsen/
ORCID: http://orcid.org/0000-0003-0330-9428
Twitter: @khinsen
---------------------------------------------------------------------