Subject: Re: egcs
To: Todd Vierling <firstname.lastname@example.org>
From: David Edelsohn <email@example.com>
Date: 08/05/1998 14:07:07
>>>>> Todd Vierling writes:
Todd> On Wed, 5 Aug 1998, Dan Jacobowitz wrote:
Todd> : > egcs 1.0.3 does not work on NetBSD/powerpc, period. egcs 1.0.3 is pretty
Todd> : > broken in code generation on anything other than i386 and sparc, basically,
Todd> Doesn't matter how fast it runs - there are some types of constructs at
Todd> higher optimization levels that will make the resulting code fall over. In
Todd> short, you can use it to generate reliable code on powerpc if you never use
Todd> -O2 or -O3.
I don't think that you can or should make this type of
generalization. egcs-1.0.3 has problems as does gcc-18.104.22.168 and
gcc-2.8.1. It all depends on how you are using the compiler and the
source code in question.
Linux had a lot of problems when switching to egcs-1.0.3 because
the EGCS optimizations have become a lot more effective and programmers
comfortable with gcc-2.7 cannot get away with as much as they used to. I
would check some of this code that you claim is broken because of code
generation that the source may be playing fast and loose with structures
and unions and types and pointers and even inlined assembly. Some people
actually were quite happy that their code began to break because it means
that GCC's optimizations have become much more powerful.