Re: Tonight's snapshot

[CaT <cat at zip dot com dot au>]

On Wed, Jul 14, 1999 at 09:09:37PM -0600, Jeffrey A Law wrote:
>   In message <19990715102333.B721@zip.com.au>you write:
>   > Erm. This sounds suspiciously close to how I've built egcs in the past
>   > and plan on building gcc-2.95. Does this mean that I wont be able to
>   > build it with -march=pentiumpro? :/ What of other s/w?
> Every compiler has bugs, that's a fact of life.  What ultimately determines
> whether or not you can compile with -march=pentiumpro is whether or not your
> code triggers that bug.
> 
> ie, just because a compiler has bugs does not make it useless -- it certainly
> sucks when you hit a codegen problem though.  But in reality every time you
> compile a program you're doing so with a compiler that has bugs.
> 
> If you are looking for some kind of info which would help you evaluate how
> your code is to trigger this bug you're kind of out of luck.  I can describe

I was generally thinking along simpler lines. I realise that the compiler
was buggy. If it wasn't it porbably wouldn't have enough features in it. ;)
What I was wondering about is will it be buggier for the ppro's now then
it was in egcs 1.1.2. I compile -everything- with -march=pentiumpro. From
glibc to egcs to X. Infact my entire system was built up from scratch
with that as an option.

So, in the end I'm wondering will things that worked before break now? :)

BTW. Thanks for keeping the development of egcs open. This was the main
thing that convinced me to use egcs over gcc (basically I trusted it
more then gcc :).

-- 
CaT (cat@zip.com.au)                       URL: http://www.zip.com.au/dev/null