> I am interested in your HDF experiences, in particular the V-set
> (which I believe is the equivalent of a FITS BINTABLE).
I wasted a lot time with Vset stuff a few years ago. It was
still in the quasi-development stages and was very unstable and
they kept changing the formats. I haven't tried anything
with it since, so I don't know if it behaves any better now.
> I am not fully in agreement with the 'self-desribing' nature of
> HDF - I have yet to see a decent program which prints out the
> contents of an HDF file in some human readable format.
> Soething like
> tsf for NEMO binary files ("i'll add an example to the nbody email
> page in another few minutes) - or any of the public domain scan-fits
> programs in the FITS tools (ftools, or related). There is some kind
> of data-dumper in HDF, but whatever I tried, never gave me anything
> I know the UIUC group is trying to standardize on something too. Who
> do you think I should contact and get some livelyhood in this discussion.
There always seemed to be somebody different over there every
time I looked. I suspect that one of his grad students, Greg
Bryan, might be the easiest to deal with. At least he'd be a good
point of contact.
> So, keep in touch, I certianly like to get more info on your data.
> Btw, the example of what you keep as data, you distibguish Nbody
> and SPH-only (like density) - so those arrays can be smaller than
> the X,Y,Z etc. arrays.
> How do you tag that? (The tipsy format has
> 3 kinds of particles - in the most general case one can have N different
> kinds of particles, each having a homogeneous set of attributes
My format is certainly not very flexible. Usually what I do is
make separate data files for the N-body and SPH particles. This
is not necessary, but just seemed easier. When I read each file,
I can tell from its array dimensions whether it's SPH or N-body particles.
I'm all in favor of a standard data format, but I don't think
it should be a big headache to use, or else no one will use it.
I really like the idea of being able to take the binary data
files directly off, say the Cray, and then read them on a
Sun or whatever, with no data conversion. That's the main
advantage of HDF, as I see it.