Re: BUG #1072: "$libdir/ascii_and_mic": No such file or directory

Yes; you're right. Sorry. Meanwhile, I also discovered, my RedHat may
originally have put a competing postgres onto the system from some RPM.

[root@hubertus root]# rpm -q postgresql
postgresql-7.2.1-5

So now I've done:
[root@hubertus root]# rpm --erase postgresql postgresql-server
postgresql-contrib php-pgsql

After having done the "rpm --erase" I cleaned the /data/postgres and
/opt/postgres-7.4.1 directories; re-created the postgres user, but initdb
still had the same fault. So I re-inserted the "absolute path"-hack;
createdb, and started psql to create function. The result from strace (when
running the CREATE FUNCTION) of the backend is:

"Q\0\0\0\237CREATE OR REPLACE FUNCTION "..., 8192, 0) =3D 160
gettimeofday({1075756990, 462661}, NULL) =3D 0
stat64("/usr/local/pgsql/lib/ascii_and_mic", 0xbfffca10) =3D -1 ENOENT (No
such file or directory)
stat64("/usr/local/pgsql/lib/ascii_and_mic.so", 0xbfffca10) =3D -1 ENOENT (=
No
such file or directory)
stat64("$libdir/ascii_and_mic", 0xbfffcad0) =3D -1 ENOENT (No such file or
directory)
write(2, "ERROR:  could not access file \"$"..., 81) =3D 81
send(8, "E\0\0\0\202SERROR\0C58P01\0Mcould not ac"..., 131, 0) =3D 131
rt_sigprocmask(SIG_SETMASK, ~[QUIT ILL TRAP ABRT BUS FPE KILL SEGV CONT STOP
SYS], NULL, 8) =3D 0
time(NULL)                              =3D 1075756990
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) =3D 0
send(8, "Z\0\0\0\5I", 6, 0)             =3D 6
recv(8,=20

It seems to search in /usr/local/pgsql -- but that directory does not exist
(at least not after the rpm --erase).

Does this give you a clue ?

Rgds
/Thomas.

=20
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]=20
Sent: Monday, February 02, 2004 9:52 PM
To: Thomas Borg Salling
Cc: 'PostgreSQL Bugs List'
Subject: Re: [BUGS] BUG #1072: "$libdir/ascii_and_mic": No such file or
directory=20

"Thomas Borg Salling" <tbs@navicon.dk> writes:
> Does this give you any clues ?

That's what I expected, but it doesn't tell me anything new.  Could you
try strace-ing the backend while you execute the CREATE FUNCTION
command?

            regards, tom lane