Re: [Texmacs-dev] RE: Texmacs segfault

[Felix Breuer]

Hello!

> > C(XX)FLAGS Flags: 
> > 
> > GCC (1) : -O0 -g -ggdb 
> > GCC (2) : -mmmx -O3 
> > GCC (3) : -march=pentium4 -mmmx -msse -msse2 -mcpu=pentium4 -O3 
> > 
> > TM (1) : -g -ggdb 
> > TM (2) : -O3 
> > TM (3) : -O3 -fexpensive-optimizations -fno-exceptions 
> > 
> > 
> > Results: 
> > 
> >         |   TM (1)   |   TM (2)   |   TM (3)   | 
> > --------+------------+------------+------------+ 
> > GCC (1) |   works    |   works    | segfaults  | 
> > --------+------------+------------+------------+ 
> > GCC (2) |   works    | segfaults  | segfaults  | 
> > --------+------------+------------+------------+ 
> > GCC (3) |   works    | segfaults  | segfaults  | 
> > --------+------------+------------+------------+ 
> 
> Thanks for this very valuable information.
> 
> Which version of guile did you use?
I used guile 1.4-r3.

> 
> Would it be possible for you to make a similar
> comparison for different versions of Guile?
I'll see if I get around to do that. These extensive recompilations take
a *lot* of time... :)

> 
> What to you mean by "an optimized g++ 3.2"?
> In other words, what did you do in order to compile g++?
I compiled gcc-3.2 using gcc-3.2 and the flags described above. I.e. I
had gcc-3.2 build three differently optimized versions of itself. The
first gcc-3.2 was not optimized, the second was medium optimized and the
thrid heavily optimized. "an optimized gcc-3.2" refers to the flags used
when building gcc-3.2 itself (with gcc.3.2). The other flags "TM (1),
(2), (3)" refer to the flags used when building TeXmacs (with the
gcc-3.2 built using "GCC (1), (2), (3)" flags).

A diagram:

            builds                  builds
GCC 3.2 ---------------> GCC 3.2 --------------> TeXmacs
         using GCC (X)            using TM (Y)

I hope this makes things clearer.

> 
> As to myself: I compiled g++ using gcc 2.95.3 without any special options
> on a RedHat 7.2 system. I compiled TeXmacs without any special options
> (i.e TM(3)) and the result seemed to work fine (I waited one hour for
> a hypothetical crash while loading and editing a few files)...
You wouldn't have to wait so long. The segfaults I experienced occurred
reliably right after every save or autosave.