Peter,
On Tue, 2 Sep 2003, Peter W. Draper wrote:
> OK, managed to get the whole thing to compile with one further change that
> I've checked in.
Phew!
> When installing AUTOASTROM the current setup ignores the
> INSTALL variable (so I just used make -e), is this intentional (i.e. do
> you intend to define the installation directory someother way)?
It's not intentional, no, though I would hope to end up specifying the
installation directory the `normal' way, with --prefix, defaulted to
/star. Obviously `make INSTALL=/star' would work here.
The medium-term plan (that is, not before ADASS) is to have autoastrom
(and astrom) configured with automake-plus-additions. That would make
supporting the --prefix option trivial. A bit of poking around
suggests that it would be easy to firstly, default that installation
prefix to /star, and secondly to add whatever other boilerplate we
required in makefiles (possibly a slightly more sophisticated
uninstall target than is in automake already, and some
datestamp-touching).
I don't really want to get enmeshed in autoastrom installation
wrinkles right now, since I'm busy getting anxious about TimeFrame and
that certainly needs to take precedence. It's probably not hard, but I
didn't think that copying this stuff into the RAL repository would be
time-consuming either....
> > The cat/CatalogInfo.h problems aren't going to go away, because the
> > library still invokes pre-standard header versions. Still, at present
> > they only provoke warnings rather than errors, so they can stay.
> > Errors associated with this, incidentally, are what stopped autoastrom
> > getting on to the Alpha CD, but I managed to work around that (I think)
> > when adding the autoconf stuff.
>
> OK, I live with these in GAIA too, although Allan Brighton has updated
> Skycat (and hence catlib) to remove this.
It should be easy to import the update into the repository. Whether I
break first, or you for GAIA, is another matter....
> OK, but now for todays error log. When I run the autoastrom script I now
> get:
>
> ------------
>
> % ./autoastrom
>
> Too many arguments for autoastrom::wmessage at ./autoastrom line 377,
> near "})"
> Too many arguments for autoastrom::wmessage at ./autoastrom line 561, near
> "})"
> Type of arg 2 to autoastrom::match_positions must be array (not hash
> dereference) at ./autoastrom line 886, near "$tfile)"
> Type of arg 3 to autoastrom::match_positions must be array (not hash
> dereference) at ./autoastrom line 886, near "$tfile)"
> Type of arg 2 to autoastrom::match_positions must be array (not hash
> dereference) at ./autoastrom line 903, near "$tfile)"
> Type of arg 3 to autoastrom::match_positions must be array (not hash
> dereference) at ./autoastrom line 903, near "$tfile)"
> Too many arguments for autoastrom::wmessage at ./autoastrom line 943, near
> "})"
> Too many arguments for autoastrom::wmessage at ./autoastrom line 1027,
> near "$iterno)"
> Execution of ./autoastrom aborted due to compilation errors.
> Undefined subroutine &main::tidy_and_shutdown called at ./autoastrom line
> 250, <DATA> line 5.
> END failed--call queue aborted at ./autoastrom line 1176, <DATA> line 5.
This looks _very_ much like Perl is reading the wrong autoastrom.pm.
The prototypes for wmessage and match_positions changed in the last
month, and those calls to wmessage and match_positions are correct for
the current prototypes, but would produce those errors when used with
the old ones.
I usually run autoastrom by setting AUTOASTROM_DIR=/tmp/... and doing
`make test-install', which just dumps the various executables and
libraries into that directory. Ah! If you're running ./autoastrom and
have AUTOASTROM_DIR defined to point somewhere under /star, then you'll
pick up the old Perl module -- is that the case?
See you,
Norman
--
---------------------------------------------------------------------------
Norman Gray http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK [log in to unmask]
|