Greetings, all
Would you like to have another go at building AST from CVS? If you
check out the current version on the dev-nxg-20031121-autoconfing branch:
cvs -d :ext:[log in to unmask]:/cvs co -r dev-nxg-20031121-autoconfing libraries/ast
Then
./bootstrap
./configure --prefix=???
make
make install
should work. `make dist' also works, and there's a tarball at
<http://www.astro.gla.ac.uk/users/norman/temp/ast-3.1-1.tar.gz> which
should build with just ./configure --prefix=xxx;make
Notes:
You need the --prefix (at the moment). The way the distribution is
rolled, and the way the checked-in sources are configured, the default
installation prefix is /export3/sun rather than /star. This makes no
sense long-term, of course, but it prompts me to point out that this
is part of a (deliberate) stickiness of the configuration. You can't
build one part of the Starlink repository without building and installing
other parts as well (for example, AST depends on messgen and sst), and
so you want the build to be configured to refer to a particular tree.
The logic of that configuration is rather intricate, and I don't think
I've got it completely sorted out yet, concerning what files should be
configured and checked in when; but it'll become clearer, and I can talk
about it at the programmers' meeting in January. Clearly (?) the
distribution tarballs should end up with a default prefix of /star.
This still doesn't build the documentation. Building the
documentation requires SST, so that has to go into the repository.
But SST requires sae_par, par_par, and others, so _they_ have to go
into the repository, too. So I'm going to work on these now, and I'd
hope to have SST's dependencies in before too long.
Part of the Plan is to have all the dependencies -- including such
`hidden' ones as AST's dependency on sae_par via SST -- sorted out and
made explicit, so that it will be easy to generate the set of
dependencies for the top-level Makefile. That should be not far off a
one-liner (albeit a biggish line...), and I'll work on that next.
Peter: you reported a problem building AST the last time I mailed
about it. I couldn't work out what the problem was from the
transcript you sent me (though I wondered if it might have something
to do with a directory `build' being mistaken for a script `build' or
something). Can you tell me if the same thing happens with this
version?
David: Though it's minus a build of the documentation, I think this is
a candidate for being merged onto the AST trunk, so you can work on
the source code again. That wouldn't stop me continuing to work on the
configuration stuff on the branch, but it would give you a useable
system to carry on development with. What do you think?
I think that's it.
See you,
Norman
--
---------------------------------------------------------------------------
Norman Gray http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK [log in to unmask]
|