On Wed, Jun 13, 2001 at 10:04:26AM +0100, Dr Russel Winder wrote:
> Zvezdan,
>
> Thanks for your suggestions about using pre-built binaries for RH6.2 on the
> RH7.1 system. Whilst this would solve the problem in some ways, it rather
> misses the point of my email.
I agree.
>
> Kayvan packages the LyX source as an SRPM which should work. Kayvan's
> packaging works, the problem is with the LyX source itself: There is a C++
> coding error in the relationship between the code in formula.C and
> lyxstring.h. It seems that either someone has changed the code in a faulty
> way between when the pre-built binaries for RH6.2 were made and the source
> that Kayvan used for the latest SRPM build or the gcc supplied with RH7.1
> (v2.96) is better at detecting C++ errors that the gcc that was used to
> build the RPMs you are using. In either case someone needs to fix the
> source.
Sorry that I didn't look more carefully at the error messages that you
gave in the e-mail. I tend to put the blame on Red Hat in such cases. I
really got annoyed with their recent blunders. Security update of 2.2.17
kernel for RH 6.2 screwed up the driver for very frequently used SCSI
card in April. Red Hat 7.0 couldn't compile its own kernel because of
the bug in their version of gcc. I had mentioned a problem with quota
over NFS with Red Hat 7.1 that I had to patch (Bugzilla bug number
42352). Other people are pointing to other problems with quota now.
I also mentioned Ghostscript issue in the previous e-mail. 6.50 compiled
fine on my home machine with the patches I made to add the support for
my printer. On 7.1 even the original RPM doesn't compile. 7.00 reports
even more serious errors in zlib library. It was simply suspicious to me
that two widely used and compiled products as zlib and LyX have
compilation errors on 7.1.
I am forwarding you below an e-mail of my colleague related to
compiler problems in Red Hat too.
A company that has an ambition to take a piece of serious enterprise
market shouldn't make such blunders IMHO (although we already know one
that took the biggest piece of market with even more serious blunders
;-)
>
> Unfortunately, I do not have the time to investigate a fix myself just now
> even though the bug itself is trivially obvious, hence the bug report via
> Kayvan to try and help the LyX Developer group.
>
> Thanks.
>
> Russel.
I understand that. And I see that Jurgen Vigna already offered much
better explanation than I could. I only wanted to point that
(Red Hat gcc != real gcc) and it might cause a serious grief.
In any case, _thank you_!
Best Regards,
Zvezdan
----- Forwarded message -----
Date: Wed, 30 May 2001 08:18:44 -0400
Subject: rh 7.1 gcc
fyi. I've for the most part had no trouble with the 7.1 gcc (it does
kernels just fine, for instance), but on the palm emulator it didn't
work. It is documented online as causing pose to cause core dumps,
and, in fact, recompiling with kgcc fixed the problem.
Since we're going to 7.1, this seems like it could be good to know.
-serge
----- End forwarded message -----
--
Zvezdan Petkovic <[EMAIL PROTECTED]>
http://www.cs.wm.edu/~zvezdan/
