Closed fhoefling closed 10 years ago
Now the species dataset is at the position /particles/atoms +-- species and in the right shape (in all 3 sample files).
The dataset /parameters/vmd_structure/indexOfSpecies is in the group vmd_structure because this dataset is at the moment unspecified by the H5MD format. Since it is VMD specific it is stored in the group vmd_structure.
What happens if another application should rely on the usage of a dataset named "/parameters/indexOfSpecies" but with different content (since the dataset is not officially specified this case could in principle happen)?
Would it be more in the spirit of the h5md documentation if the prefix "vmd_" would be prepended to all datasets which are currently located in the group " /parameters/vmd_structure/"?
/parameters is the right place to store application-specific data. You are completely free in what is stored there and how. It is a good idea to put these data in an application-specific subgroup (what you did), although this is not explicitly required by the H5MD specification. The subgroup should avoid potential conflicts with other programs.
My only concern was that the shape of "indexOfSpecies"
in samples/half_complete_vmd_structure.h5
is still Dataset {3/Inf, 1}
, just as for "species" before the fix. In full_vmd_structure.h5
this has been fixed.
Actually, what is still missing in the files are the metadata in /h5md
and a box specification in /particles/atoms/box
. These fields are apparently not evaluated by the VMD reader.
Ok, there is a new issue for the box data.
The species data in the sample files is stored in a non-H5MD-compliant format:
But a time-independent H5MD element is stored as a dataset itself, not a group:
The same applies to /parameters/vmd_structure/indexOfSpecies.