This is the mail archive of the
mailing list for the GCC project.
Re: finding strict aliasing problems
Jakub Jelinek wrote:
> From my experience (gcc-2.96-RH defaults to strict-aliasing) the vast
> majority of sources we've been compiling is ok, there were a couple of bugs
> (like in X11) which have been fixed since then and ATM I'm aware of just 2
> packages which have not been fixed yet (clapack and one more).
I must say that - in retrospect - I find this a very courageous move by
Red Hat. It might easily have been responsible for 80-90 % of the
complaints about (the C-side) of the compiler suite at the time Red Hat
Linux 7.0 was released (in September last year).
I recall quite well the avalanche of "bug reports" dealing with well
known packages that turned out to be strict aliasing problems in 1999
(which was the reason to turn it off in 2.95.2).
If this really resulted in there now being only two widely used packages
showing problems in this regard, I must say I'm much impressed (not in
the least by all the work Jakub and other Red Hatians must have expended
on tracking down these bug reports and suggesting remedies).
Toon Moene - mailto:email@example.com - phoneto: +31 346 214290
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
Join GNU Fortran 95: http://g95.sourceforge.net/ (under construction)