At 10:46 AM +0100 7/27/00, Patrick Welche wrote:
>The package probably slaps the -Wl,-R/usr/pkg/lib in. The point of the
>above is "use LD_LIBRARY_PATH or ldconfig" isn't a solution for us lot,
>the -Wl,-R (or -rpath) is needed (say configure --not-Debian or something ;)
Shouldn't there be an option to disable shared libraries altogether?
I don't see that they buy us much in most real scenarios. Every copy
of the postmaster already shares the same complete executable image
by default. (Actually that bit of smarts was built into old Unix
Version 6 or earlier.) That's actually a bigger savings and you get
it without the run-time overhead of shared libraries. If we have
large numbers of clients on the same machine they probably share the
same executable image as well.
> > > > > Third problem: well actually the regression tests seem to
> > >work, mostly. ;-)
> > >
> > >Where they the parallel regression tests? Does "unlimit maxproc" help? (I
> > >usually forget to do this and maxproc=80 isn't enough for me)
> >
> > Tried it. No change. I did have to stop the running postmaster to
> > do a gmake runcheck or it fails because it can't get enough shared
> > memory. I'm running an unoptimized generic kernel.
>
>Fair enough - same here. The maxproc business for me would be that a new
>backend can't be forked for a given test => it doesn't output anything
>other than an error message => the test fails. What sort of errors are
>you getting?
I emailed the results files for the failed tests to Thomas L. (but
not Tom L.). If anyone else wants copies just ask.
Signature failed Preliminary Design Review.
Feasibility of a new signature is currently being evaluated.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu