[fitsbits] [Fwd: FITS to HDF5 converter]

[Manually forwarded to FITSBITS because the sci.astro.fit-FITSBIT gateway
appears to be broken]

-------- Original Message --------
Subject: FITS to HDF5 converter
Date: Mon, 23 Jan 2006 17:38:11 -0600
From: Pedro Vicente Nunes
Organization: Usenet @ UIUC - http://www.killfile.org/usenet/
Newsgroups: sci.astro.fits

The HDF group at NCSA is implementing a FITS to HDF5 converter. HDF5 is a
general purpose library and file format for storing scientific data.

We wrote an RFC (request for comments) of the software, that can be found
here

http://hdf.ncsa.uiuc.edu/RFC/fits2h5/fits2h5.htm

one thing that we would like to do is to ask for the FITS community for
suggestions about the preferred way to write the HDF5 file

your comments would be highly appreciated

thank you

for more information about HDF5, see

http://hdf.ncsa.uiuc.edu/

On Thu, 26 Jan 2006, William Pence wrote:
[Manually forwarded to FITSBITS because the sci.astro.fit-FITSBIT gateway

From: Pedro Vicente Nunes

http://hdf.ncsa.uiuc.edu/RFC/fits2h5/fits2h5.htm
your comments would be highly appreciated

I've seen your request when Bill forwarded it on the mailing list. I'm
not at all an HDF user, but I looked at your page and have the following
(minor) comments.

- You do not make reference to the RFC which describes the FITS format
(RFC4047), nor to other references for the FITS standard (see such
RFC or maybe Bill can provide pointers)

- in your brief descriptions of FITS you mention image extensions and
binary and ascii table extensions. You do not mention the random
group extension (being this now deprecated but for specialized uses,
it may make sense to ignore it)

- in "Naming HDF objects" you propose to use names like HDU_n. There
are however cases when extensions have a name (EXTNAME), sometimes
this is explanatory for the user.

- you translate keywords into your attributes, as compound of kwd name,
kwd value and comments. Of course it's up to you, I do not know how
you use such attributes. But why not NAMING them with the keyword name?

For instance when I read data files (FITS or homegrown) into IDL
structures, I may map things like this (for simplicity I consider a file
with just the primary HDU or a single table extension) :

- file maps to structure "a"
- a primary DU maps to an element a.data which is array
- an extension DU maps to an element a.data which in turns is a
structure with as many elements as table columns, with the column
name e.g. a.data.mycolname

- all header keywords match to string elements e.g. a.bitpix,
a.naxis, a.naxis1 etc. named according to the kwd name ... in a
variation some keywords are interpreted as a numeric type, but
still the element a.keyname is named according to the kwd name

- I also omit keywords which are irrelevant (e.g. SIMPLE is fixed),
or comments. I am instead forced to change names for keywords
which may occur more than once (like HISTORY) using sequential
names e.g. a.h00001 a.h00002

--
----------------------------------------------------------------------
is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.