Re: 9.1 (git head) does not compile using --with-libedit-preferred on Ubuntu 10.10

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Mark Kirkwood <mark(dot)kirkwood(at)catalyst(dot)net(dot)nz>
Cc: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: 9.1 (git head) does not compile using --with-libedit-preferred on Ubuntu 10.10
Date: 2011-02-16 01:54:45
Message-ID: 875.1297821285@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Mark Kirkwood <mark(dot)kirkwood(at)catalyst(dot)net(dot)nz> writes:
> Since libedit is getting some attention right now, I figured I'd try
> using building with it instead of readline. configuring using:

> ./configure --prefix=/usr/local/pgsql/9.1 --enable-debug
> --enable-cassert --with-libedit-preferred

> I get this linking postgres:

> postmaster/postmaster.o: In function `PostmasterMain':
> /home/postgres/develop/c/postgresql/src/backend/postmaster/postmaster.c:755:
> undefined reference to `optreset'
> tcop/postgres.o: In function `process_postgres_switches':
> /home/postgres/develop/c/postgresql/src/backend/tcop/postgres.c:3457:
> undefined reference to `optreset'
> utils/misc/ps_status.o: In function `set_ps_display':
> /home/postgres/develop/c/postgresql/src/backend/utils/misc/ps_status.c:314:
> undefined reference to `setproctitle'

> What seems to be going on is that libedit defines optreset, which then
> messes up the various other bits of logic in utils/misc/ps_status.c so
> it thinks it has setproctitle on Linux.

It's pretty hard to see how those two things would be related. I think
more likely libedit is providing a function named setproctitle, which
seems like a rather stupid thing for them to have done.

The optreset thing is probably the same ilk of issue, ie, libedit
providing such a symbol and configure then being fooled into expecting
it to be present in the standard libraries.

We could possibly fix this by delaying the test for libedit till after
we've tested for unrelated stuff, but that seems pretty klugy.

> I'm not sure how important this is, given that even if I (bodge) these
> errors away, I merely confirm that libedit multi-byte input is busted in
> the version shipped with Ubuntu 10.10.

There's that too :-(

regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2011-02-16 01:59:18 Re: updated patch for foreach stmt
Previous Message Bruce Momjian 2011-02-16 01:53:03 Re: pgsql: Adjust pg_upgrade error message, array freeing, and add error ch