Michael J Schout wrote:
> We would definately beta test 7.1 beta releases on our test machines if RPMS
> were made available. However, if rpms are not made available, its unlikely
> that anyone around here will get time to build the sources from scratch. So
> you can count me as one beta tester that you would have if you made RPMS of the
> betas.
Your offer is appreciated, and will most definitely be taken :-).
I'm experiencing some degree of difficulty with the build -- mostly due
to some reorg in the Perl and Python clients, but also some main Make
framework as well, thanks to the RPM build-root environment. The
compiles I have done outside the RPM environment have worked and
installed (as source installs) very cleanly -- but the RPM build-root
environment is rather different -- installing files to a location where
they won't actually be installed to :-).
Let me explain: so that RPM builders don't accidentally trash their
systems during building, RPM includes a 'build-root' mechanism that
allows a fake root for the build install to be used instead of the real
root. Think 'chroot-lite'. This build-root is not enforced anywhere
except by the spec file build and install sections. This also allows
RPMs to be built for root installation by a non-root user, which
provides an extra layer of filesystem protection.
So, files get installed to this build-root, for eventual real
installation on the real root when the RPM is actually installed.
However, there are some hard-coded paths left in the build, and the perl
client is being difficult, and odbcinst is going to the REAL /usr/etc
instead of $RPM_BUILD_ROOT/etc.... I have lots of combing to do. In
many ways 7.1 is an easier build -- but not in this regard. But I
consider this an RPM issue and not a PostgreSQL tarball issue, meaning,
while I will be developing patches for the RPM build, I won't expect
those to be integrated into the main tarball.
So, I'm plugging at it....
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11