Linking problems with gcc 3.4.2 and glibc on 5.3-stable

[Chris]

Well it started out with me getting a complaint from a user on my
server saying he couldnt compile psybnc since I upgraded to 5.3 I told
him I would investigate when I get time, I then found out later that
ezbounce also failed (i use ezbounce) on 5.3, but my binary which was
backed up from when I compiled on 5.2.1 works fine.  I also heard from
many friends on irc that psybnc doesnt work with 5.3 so obviously the
problem is wide spread and it was one reason why a lot of shell
companies were not upgrading from 4.x yet.  I then decided to try
psybnc myself and this is what happens using gcc 3.4.2 bundled with
5.3.

Initializing bouncer compilation
[*] Running Conversion Tool for older psyBNC Data.
tools/convconf.c: In function `cofile':
tools/convconf.c:81: error: label at end of compound statement
*** Error code 1

Stop in /usr/home/chrysalis/psybnc.

I can duplicate this problem every single time on any 5.3 box it has a
100% failure rate so I would be interested to know how psybnc worked
for you, however if I install gcc 3.3.6 and then edit the top line in
Makefile to cc = gcc33 then it compiles fine. I will as you suggest
try a later gcc in the 3.4.x branch as there is newer version then
what is bundled in 5.3 and I will report back. I havent test ezbounce
with 3.3.6 yet.

Chris

On Wed, 26 Jan 2005 21:06:26 +0100, Björn König <bkoenig at cs.tu-berlin.de> wrote:
> Chris schrieb:
> 
> >there is also a bug which prevents compiling of ezbounce and psybnc, I
> >wonder how much testing went into the new gcc compiler or was it just
> >bundled into the release, considering this is the reccommended release
> >for production server's now this is a bad oversight, I think the gcc
> >should be changed in the next patch level back to 3.3.6.
> >
> >
> I compiled psybnc already a few times with FreeBSD 5.3 successfully. If
> there is a problem then please tell it.
> 
> Regards Björn
> 
>