36 messages in com.mysql.lists.perlRE: same problem with symbol net_buff...
FromSent OnAttachments
kryp...@yahoo.com31 Jul 2003 13:08 
Michael Koch01 Aug 2003 01:57 
Thomas Whitney04 Aug 2003 19:52 
Rudy Lippan05 Aug 2003 06:30 
Thomas Whitney05 Aug 2003 07:08 
Thomas Whitney05 Aug 2003 07:47 
Thomas Whitney05 Aug 2003 07:53 
Rudy Lippan05 Aug 2003 07:56 
Karen Ventura05 Aug 2003 09:40 
Thomas Whitney05 Aug 2003 10:04 
Michael Koch05 Aug 2003 10:06 
Thomas Whitney05 Aug 2003 10:23 
Karen Ventura05 Aug 2003 10:29 
Rudy Lippan05 Aug 2003 10:57 
Karen Ventura05 Aug 2003 11:37 
kryptoer05 Aug 2003 13:27 
Rudy Lippan05 Aug 2003 14:11 
Karen Ventura05 Aug 2003 14:57 
Thomas Whitney05 Aug 2003 15:25 
Rudy Lippan05 Aug 2003 16:05 
Thomas Whitney05 Aug 2003 16:47 
Jim Cromie06 Aug 2003 00:27 
Thomas Whitney06 Aug 2003 04:30 
Rudy Lippan06 Aug 2003 10:04 
Karen Ventura06 Aug 2003 10:48 
Thomas Whitney06 Aug 2003 13:51 
Rudy Lippan06 Aug 2003 20:20 
Rudy Lippan06 Aug 2003 23:50 
Thomas Whitney07 Aug 2003 05:35 
Michael Koch07 Aug 2003 06:49 
Rudy Lippan07 Aug 2003 09:24 
Rudy Lippan07 Aug 2003 09:48 
Michael Koch08 Aug 2003 02:25 
Michael Koch08 Aug 2003 03:10 
Michael Koch08 Aug 2003 05:43 
Michael Koch08 Aug 2003 09:33 
Subject:RE: same problem with symbol net_buffer_length not found with 'make test' for DBD::mysql
From:Karen Ventura (kkn@tln.lib.mi.us)
Date:08/06/2003 10:48:35 AM
List:com.mysql.lists.perl

Rudy,

Your clarified "perl Makefile.pl" command has made all the difference. Using that to begin with, I no longer have any problems with DBD. So the key for me was to install as:

perl Makefile.PL --cflags=-I/usr/local/mysql/include/mysql --libs='-L/usr/local/m ysql/lib/mysql -lmysqlclient -lcrypt -lnsl -lm' make make test make install

Then, everything was great. Thanks again!! :)

Karen

-----Original Message----- From: Rudy Lippan [mailto:rlip@remotelinux.com] Sent: Wednesday, August 06, 2003 1:05 PM To: Jochen Wiedmann Cc: Thomas Whitney; kkn@tln.lib.mi.us; Michael Koch; pe@lists.mysql.com Subject: RE: same problem with symbol net_buffer_length not found with 'make test' for DBD::mysql

On Wed, 6 Aug 2003, Jochen Wiedmann wrote:

Three questions:

1.) If the problem *now* is, that libmysqlclient.so wasn't found: What had been the problem before?

Before the problem was that mysql.so was not being linked against libmysqlclient, so it had undefined symbols but did not know that it needed to load up libmysqlclient.so to resolve them. Which is why LD_PRELOAD worked.

2.) Related to that: What did the linker statement

cc -shared -L/usr/local/mysql/lib/mysql -L/usr/local/lib dbdimp.o \ mysql.o -o blib/arch/auto/DBD/mysql/mysql.so -lmysqlclient

effectively change? IMO this was right as in some other mails?

They linker command that I somehow ended up with was missing the '-lmysqlclient' (or put it before mysql.so -- I dont' remember) I still don't know exactly what was/is going on here and what causes this...

Wait.... Here is something:

perl Makefile.PL --cflags=-I/usr/local/mysql/include/mysql --libs=-L/usr/local/mysql/lib/mysql -lmysqlclient -lcrypt -lnsl -lm

Looses -lmysqlclient -lcrypt &al. the command should be:

perl Makefile.PL --cflags=-I/usr/local/mysql/include/mysql --libs='-L/usr/local/mysql/lib/mysql -lmysqlclient -lcrypt -lnsl -lm'

3.) Doesn't the above again depend on the presence of the environment variable? As was the case for LD_PRELOAD before? IMO a real solution should avoid the necessity of a variable.

LD_LIBRARY_PATH just tells the dynamic linker where to search for its shared object. If you don't want to use LD_LIBRARY_PATH, you can modifiy /etc/ld.so.conf and add the path to where the shared object lives -- which is the standard why of doing things if you have root access to the system.