Re: Fwd: Survey: utility of packaging Hipparcos star catalog?
On 2008.04.11 10:24:12 CEST, Fabien Chéreau wrote:
Hi Johannes,
This is a survey for our star expert!
Fabien
---------- Forwarded message ----------
From: Kevin B. McCarty <kmccarty@debian.org>
Date: Fri, Apr 11, 2008 at 1:16 AM
Subject: Survey: utility of packaging Hipparcos star catalog?
To: debian-science@lists.debian.org
Cc: celestia@packages.debian.org,
education-astronomy@packages.debian.org, kstars@packages.debian.org,
openuniverse@packages.debian.org, starplot@packages.debian.org,
stellarium@packages.debian.org, xorsa@packages.debian.org,
stardata-common-devel@lists.alioth.debian.org
Hi,
you're receiving this email because you are either subscribed to
debian-science or you maintain a package that was tagged
"field::astronomy" and looked like it might be relevant.
I'm thinking of packaging the Hipparcos star catalog [1] for
Debian. It
is already included in the stellarium-data, celestia-common and
kstars-data packages [2,3,4] in some form, but it is in a binary
format
in the first two of these, and it seems to me that it might be
useful to
have the untouched original version available in a more
general-purpose
package. Since users can easily download it themselves from [1], the
main point would be for use in Depends or Build-Depends(-Indep).
Hence
this survey. If you reply, please do so to
debian-science@lists.debian.org.
[1] http://cdsarc.u-strasbg.fr/viz-bin/Cat?I/239
[2] http://packages.debian.org/sid/stellarium-data
[3] http://packages.debian.org/sid/celestia-common
[4] http://packages.debian.org/sid/kstars-data
* Question 0: Does (or can) the astronomy software you maintain make
use
of the Hipparcos data in any way? (If not, there's no need to reply,
although you can of course weigh in anyway. Note, I consider the
much
larger Tycho catalog [5] to be separate and have no plan to package
it.)
Stellarium itself cannot access the Hipparcos data directly, although
all Hipparcos stars are required.
[5] ftp://cdsarc.u-strasbg.fr/pub/cats/I/239/tyc_main.dat.gz
* Question 1: Can your software make use of the Hipparcos catalog
exactly in its original format (the original version of the main
file is
at [6] for instance -- assume it would be gunzipped), with no
transformation to a different format?
[6] ftp://cdsarc.u-strasbg.fr/pub/cats/I/239/hip_main.dat.gz
The stellarium program cannot use the hipparcos catalogue in its
original format.
* Question 2: If "no" to question 1, then is it possible to recreate
the
file(s) in the format used by your software from the raw Hipparcos
data
programmatically? For instance, if upstream has written a script
that
can automatically convert the hip_main.dat file to
some_file(s)_your_program_reads.txt, perhaps with the help of some
additional auxiliary input files.
(I would assume that stellarium and celestia upstreams must have some
way of generating /usr/share/stellarium/stars/default/*.cat and
/usr/share/celestia/data/stars.dat, respectively, from the Hipparcos
files. Unfortunately it looks like the files in kstars-data were
edited
by hand to a large degree.)
For stellarium the *.cat files are generated in 3 steps by 3 seperate
programs that are distributed in the util directory. Following input is
required:
1) hip_main.dat,h_dm_com.dat (errata.htx is also taken into account)
2) Tycho
3) NOMAD (this is about 100GB)
* Question 3a: If "yes" to either question 1 or 2, would your
software
or transformative script/program be able to cope with the
hip_main.dat
file being split into several smaller files in the Debian-packaged
version? (I might do this in order to permit people who are low on
disk
space to install only the part of the catalog including, say, stars
within 30 parsecs of Earth.)
hip_main.dat is such a tiny little file that I see no sense in
splitting. Also note that the stars in Hipparcos are less then 0.1% of
all the stars in stellarium. And no, the transformation program
requires not some splitted files, but hip_main.dat,h_dm_com.dat from
Hipparcos (plus Tycho and NOMAD).
* Question 3b: Does your answer to 3a depend on the exact way in
which
the hip_main.dat file is split up (i.e., which stars go into which
piece)?
* Question 4: If "yes" to question 2, would it make more sense for
you
to (a) Build-Depend upon a Hipparcos catalog and generate your own
foo-data Debian package, or (b) do the transformation on the user's
computer in a postinst script at install-time? The former has the
downside of filling ftp.debian.org with multiple copies of the
Hipparcos
catalog in various formats. The latter has the disadvantage of
making
the local admin wait, possibly for a long time, while the
package-specific version of the catalog is generated. There is
support
for the latter via the stardata-common package [7,8].
[7] http://packages.debian.org/sid/stardata-common
[8]
http://alioth.debian.org/plugins/scmcvs/cvsweb.php/stardata-common/doc/policy.txt?rev=1.1;content-type=text%2Fplain;cvsroot=stardata-common;only_with_tag=HEAD
I hope it has become clear by now that the Hipparcos catalog is far
away from sufficient, when it comes to the amount of stars presented by
contemporary astronomy programs. Even the stars[0123].cat that come
with the basic distribution of stellarium, contain more than 4 times as
many stars as the hipparcos catalog.
You can of course answer (c) "neither" to this question if you want
to
keep shipping the files that upstream provides in both your source
and
binary packages -- the Hipparcos catalog is basically public-domain
so
that's perfectly OK.
Thanks very much for your time.
best regards,
Yours,
Johannes Gajdosik (stellarium)
--
Kevin B. McCarty <kmccarty@gmail.com>
WWW: http://www.starplot.org/
WWW: http://people.debian.org/~kmccarty/
GPG: public key ID 4F83C751
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFH/p/5fYxAIk+Dx1ERAuaSAKCySxcy5QsOXX87aagzCFKY5snFMwCeOGan
h0SY9WUO3bDoBaRWr4DSq/0=
=EJDS
-----END PGP SIGNATURE-----
Reply to: