compiling glib-2.3.1 with gcc-3.2.2 --enable-oldest-api=2.2.5 problem

mclinden at informed.net mclinden at informed.net
Fri Feb 21 08:46:59 PST 2003


First of all, I didn't mean to knock the toolchain and this is not a flame 
but simply a comment. 

I am a big fan of LFS which has allowed me to basically build CD-based 
installs that take 10% of the time as vendor distro installs. In fact, it 
has been so stable for me that I am able to guarantee a supported Oracle 
install on a configuration that Oracle wouldn't touch.

But the evolution of LFS is occurring in the context of developments in 
GLIBC, GCC and binutils, as well as the Linux kernel, all of which can 
confound the system builder, especially when Pure LFS Hints require a CVS 
of GCC. 

For me, the test of the success of the deployment is that you can actually 
run a production system on it not simply that it boots. The reality is 
that there are some legacy drivers, libraries, apps and the like for which 
source code is not available and for which portability is an issue for 
those of us who use LFS in production. 

Note that I don't believe that this is a problem for the LFS development 
team. My original query was simply to determine whether anyone had seen 
this type of problem before.

But more specifically, the issue is the fact that glibc-2.3 no longer 
includes the __ctype_ operators (except when you enable the abi 
compatibility)  and the link editor complains when you try to relink 
legacy (Oracle) apps. It doesn't appear to be a problem insofar as the 
apps run, but the errors cause automated installs (like Oracle Installer 
and ./configure) to fail. I can get around problems like this using such 
nice flags as -Wno-deprecated and -Wl,--allow-shlib-undefined but as a 
purist, I prefer to fix warnings rather than ignore them.

However, it appears to be a moot point since the problem (I just, now, 
discovered) lies squarely with libc developers. Apparently the 
scripts/abi-versions.awk file generates a bad header file when 
--enable-oldest-abi=2.2.5 or --enable-oldest-abi=2.2 but works fine for 
--enable-oldest-abi=2.0.

Go figure.

Sorry for the waste of bandwidth. Maybe I'll write a Hints so I don't 
forget this.

Sean







Gerard Beekmans <gerard at linuxfromscratch.org>
Sent by: lfs-dev-bounce at linuxfromscratch.org
02/21/2003 12:38 PM
Please respond to lfs-dev
 
        To:     lfs-dev at linuxfromscratch.org
        cc: 
        Subject:        Re: compiling glib-2.3.1 with gcc-3.2.2 
--enable-oldest-api=2.2.5 problem


On February 21, 2003 08:33 am, mclinden at informed.net wrote:
> experience with LFS versions 2 and 3 but 4 has been problematic (I 
suspect
> that the toolchain issue has something to do with it).

What kind of problems have you been experiencing exactly? Yes, the 
toolchain 
build in the 4.x series leaves some to desired, but fundamentally it's not 
a 
whole lot different from the LFS-3.x series. Sure a bit different, but any 

major problems are more likely to stem from newer versions and their own 
bugs 
than from the build method. The current method is fragile, but if done 
properly it's not entirely brain dead either.


-- 
Gerard Beekmans
www.linuxfromscratch.org

-*- If Linux doesn't have the solution, you have the wrong problem -*-
-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list