I appreciate all the responses and the ideas generated by my request
which I sent out on the netCDF network:
Date: Tue, 18 Feb 1992 23:53:45 EST
From: mchance (Morrell Chance)
The netCDF file system seems quite elaborate, with excellent features,
and seems to have a popular following. As someone new to these so
called standards, I am not quite sure which to choose. There seem to
be other interfaces around which have some claims to being 'standard'
also. Like, for example, the NCSA HDF system. And what about others
like WEB and BASIS? There may also be others. Has anyone compared
these against each other? Is there any sense in such a comparison? I
would like to hear more from the experts to help me get a start on
this. Any information woiuld be helpful.
I had a rather modest motivation behind my request for a
standard. We have a number of codes here at the Plasma Physics
Laboratory which are or can be linked together, usually through data
files. They are usually written by a variety of authors and even
though we all are at most a few offices or so away from each other
linking codes can still be problematic through misinterpretations,
idiosyncrasies, etc. Moreover, other institutions have codes with
which ours communicate, and it's sometimes a struggle to verify that
the units, formats, etc. are identical. These codes may also cross a
variety of machine platforms. Obviously, a standard would be useful
here. On top of this, some of these codes also communicate with data
garnered from experiments, and I see that there ars also standards for
amassing data through instrumentation (the ADISS project). Another
thing which we do is display results, both textually and graphically,
across different machines and operating systems.
The two dominant standards seems to be netCDF and HDF and I
see that at least one group may incorporate both of these. I see also
from Glenn Davis's and Russ Rew's article comparing these that NCSA
may merge these two standards. This may be important since I gather
from reading the article that HDF has the big advantage (for me) of
its raster-image manipulation. I wonder when this merger is going to
take place?
I don't yet forsee my problems as being as esoteric as the
recent one on the network about the storage of time slices. I would
like something which is actively supported and maintained, and easy to
implement and sell. I have to try these out for myself or talk
directly to users. (I have used HDF some).
Thanks again for the info and discussion. Any more suggestions
would be most welcome.
Morrell Chance
PPPL
E-mail: chance@xxxxxxxxxxxxxxx
or: chance@xxxxxxxxxxxxx
