Is there i faq item or maybe some other way that we can tell users
that --disable-rpath should be used when checking regressiontests if you
have an install already on the box?
The rpath defaults to the installdir even in the temporary install that
is made when doing "gmake check", and that confuses some programs (like
pg_encoding) because it loads the installed libpq. In my case that's an
old install running in production state.
There are tools that set the rpath on an already linked executable, but
i dunno if that feels appropriate. But maybe an warning in "gmake check"
would be nice.
Magnus
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-Programmer/Networker [|] Magnus Naeslund
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-