Dave Page wrote:
> Christopher Browne wrote:
> > cbbrowne at dba2:Slony-I/CMD/slony1-HEAD/tests> cat
> > /tmp/slony-regress.c12207/slonik.log
> > <stdin>:4: PGRES_FATAL_ERROR load '$libdir/slony1_funcs';  - ERROR:
> > could not load library "/opt/OXRS/dbs/pgsql82/lib/slony1_funcs.so":
> > /opt/OXRS/dbs/pgsql82/lib/slony1_funcs.so: undefined symbol:
> SET_VARSIZE
> > <stdin>:4: Error: the extension for the Slony-I C functions cannot be
> > loaded in database 'dbname=slonyregress1 host=localhost user=cbbrowne
> > port=5882'<stdin>:4: ERROR: no admin conninfo for node 134594856
> >
> > Note: this problem is consistent across both branches (e.g. - HEAD
> and 1.2).
> >
> > I suspect that this breaks something about support of older versions.
> > At some point, that may be acceptable for CVS HEAD, but I rather think
> > it won't be OK for 1.2 :-).
> >
> > I don't have time to explore that this afternoon; hopefully our
> > timezones differ sufficiently appropriately that you either see this
> > tonight or tomorrow morning.
>
> Sorry - thinko on my part combined with lack of testing on 8.2 (too
> intent on making 8.3 work :-( )
>
> Updated patches attached, each minus the extra = char that they were
> previously sporting!
>
> Regards, Dave
>
Sorry it took a while to test this; I ran the test:

   ./run_test.sh test1

against both the 1.2 and HEAD branches,

against Debian/testing

against PostgreSQL versions 7.4.5, 8.2.3, and today's CVS HEAD (e.g. -
8.3 'real, real beta').

I'll want to run against 8.0 and 8.1 as well, at some point; running 6
regression tests seems enough to indicate that it can't be too terribly
broken :-), especially when we have successful test runs on every other
major version, and it worked on both older and newer versions...

I'll be checking the changes into CVS for both branches shortly.

Thanks, Dave!