Re: Question on pg_dump

From: Michael Brusser <michael(at)synchronicity(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Question on pg_dump
Date: 2004-02-27 15:18:20
Message-ID: DEEIJKLFNJGBEMBLBAHCAEANEFAA.michael@synchronicity.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

I first tried to take care of the problem by removing "-L $libpath"
from the arg list passed to createlang. This worked in a way that
probin in pg_proc had value "$libdir/plpgsql".

Later it turned out the embedded library path was used, and install
failed when there was no access to the build environment.

Now I put the "-L $libpath" argument back in place, then I update
pg_proc with the dynamic value. This works fine, but such approach
looks too intrusive. I think I also tried to set env LD_LIBRARY_PATH,
which by some reason did not help.

Is there a better way to handle this and avoid updating pg_proc?

Thank you.

========================================================
> Michael Brusser <michael(at)synchronicity(dot)com> writes:
> > I'm running Postgres v.7.3.4.
> > In my database dump file I see this:
>
> > CREATE FUNCTION plpgsql_call_handler () RETURNS language_handler
> > AS '/home/tmichael/build/relipg21/syncinc/lib.sol2/plpgsql',
> > 'plpgsql_call_handler'
> > LANGUAGE c;
>
> > The hardcoded library path may become an obstacle when loading
> > data into a different server. Is there a way to avoid this?
>
> The preferred way to write it nowadays is '$libdir/plpgsql', but
> you evidently have a legacy value embedded in your pg_proc table.
> pg_dump will not second-guess this, and so the old full-path
> approach will persist over dump/reloads until you do something about it.
>
> I'd suggest editing the dump file before you reload, or even manually
> updating pg_proc.probin for this function entry so that future dumps
> are right.
>
> regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Michael Meskes 2004-02-27 15:24:31 Re: Thread safe connection-name mapping in ECPG. Is it
Previous Message Greg Stark 2004-02-27 14:57:24 Re: [pgsql-www] Collaboration Tool Proposal