David, and all.
On Fri, 17 Oct 2003, Giaretta, DL (David) wrote:
> I'd be grateful if all those who were at ADASS and the IVOA could give some
> thought to the implications of what was seen/heard for Starlink s/w.
Data Models are in the air, and if we can produce some _practical_
application of those ideas it would look very good indeed, I believe.
After talking to folk in Strasbourg, I'm less optimistic that RDF
Will Save The World, but somewhat clearer about just what a practical
application of the ideas would be. I have some thoughts about that,
which I'll work on.
The other thing was our build procedures. After talking with Brian
Thomas, and Martin Hill of AstroGrid, I'm more convinced than ever
that
We will have massive difficulty getting anyone to use our classic
applications or libraries while ./mk exists.
If folk can't type ./configure;make they immediately and automatically
think the software is wierd/old/quixotic/unstable/unmaintained,
matteradamn how much you say otherwise in the README. I know that
there are a couple of counterexamples such as Bill Joye, AST and DS9,
but the fact that there are so few for an obviously superior library
should tell us something.
If we're meeting face-to-face again soon, we should thrash something
out then. Does anyone else think this is this urgent?
Overall I think this meeting was even more successful than the IVOA
one. We participated a _lot_, and have probably made Starlink as well
known, and well thought-of, internationally as it has ever been.
Things I need/want to do next:
* TimeFrame -- quite a lot to do, including finishing the functionality,
adding the documentation, and reading Arnold's STC stuff.
* CVS -- no hassle.
* Agitate about the build procedures. I might be able to persuade
DavidB that autoconfing AST might be useful (David?); Peter challenged
me to try the trick with KAPPA (I'm not getting at David's software,
but AST is what I'm currently working on, and Peter claimed that if
it worked with KAPPA he'd believe it would work generally). If we
automake things, then it should be pretty easy to roll
distribution tarballs that look like folk expect, so they can
_get_ our software without browsing it in the CVSweb archive, and
without going through the odd-looking software store. In other
words (again) a more contemprary build procedure can help us have
a more contemporary distribution look-and-feel.
* Work on some practical RDF application.
Generally, I think it's important that we carry on participating in
IVOA groups (I know we weren't planning to stop). I'm in the UCD group,
plan to keep sticking my oar into the DM group, and have just sent Ray a
pile of comments on the Registry Metadata proposals. That keeps up our
visibility, and thus makes it easier for folk to consider incorporating
our software into their systems and applications -- we're no longer some
odd UK project no-one's ever heard of, but manifest geniuses, whose
software they should get their paws on as soon as possible. Which is
why a normal-looking distribution procedure is important, by the way.
See yez,
Norman
--
---------------------------------------------------------------------------
Norman Gray http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK [log in to unmask]
|