Re: Bug in egcs-2.90.29 (Linux systems)

[Cautaerts Vincent <vincent at comf5 dot comm dot eng dot osaka-u dot ac dot jp>]

Hello,

Of course, I made a fool of myself...

Horst von Brand <vonbrand@sleipnir.valparaiso.cl> said:
-> Cautaerts Vincent <vincent@comf5.comm.eng.osaka-u.ac.jp> said:
-> > 2. The bug does no apper if any of the following is done:
-> > 2.1. Run the preprocessor separately
-> >   gcc -E Debug.cc -O2 >Debug2.c; gcc -c Debug2.c -Wall
-> 
-> In that case no optimization is done (the last gcc call is the one really
-> doing the compiling).  Also, you are compiling the resulting file as C, not
-> C++. This is very weird, as your testcase definitely _isn't_ C. I get loads
-> of errors that way.

Effectively, I made a mistake when recopying the line... (changing g++
to gcc), but I forgot effectively to include the "-O2" for the
optimization itself. Doing it, the bug appears also (hopefully, it is
not some strange interaction between the preprocessor and the
compiler)

-> > 4. The bug does not happen on FreeBSD
-> 
-> Same compiler? Same libc et al?
No, I forgot to check (sorry to have waisted your time...). It is
a simple gcc there (gcc version 2.7.2.1).

-> BTW, here with linux-2.1.131ac13, egcs-19981213, glibc-2.0.108 it compiles
-> fine. It would seem that the bug is fixed in the latest snapshot. You might
-> try with egcs-1.1.1 (I understand the relevant RPMs are in the rawhide
-> distribution), at least here it compiles without complaints using
-> self-compiled egcs-1.1.1 C++ compiler.

Thanks for the info. Effectively, it solved the problem.

BTW, thanks to all of you for making a free compiler of such a quality.

-- 
Vincent Cautaerts
Quote of the day/week/month:
  "The fundamental principle of science, the definition almost, is this: the
sole test of the validity of any idea is experiment." -- Richard P. Feynman