[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Chicken-users] llvm-gcc / clang
From: |
F. Wittenberger |
Subject: |
Re: [Chicken-users] llvm-gcc / clang |
Date: |
Fri, 03 Dec 2010 21:27:02 +0100 |
Am Freitag, den 03.12.2010, 03:30 -0500 schrieb Felix:
> From: Thomas Chust <address@hidden>
> Subject: Re: [Chicken-users] llvm-gcc / clang
> Date: Thu, 2 Dec 2010 16:16:21 +0100
>
> > 2010/12/1 Felix <address@hidden>:
> >> [...]
> >> Oh, and clang gave me stupid warnings that where actually wrong
> >> (an "x == x" comparison of floats to detect NaN, which is IMHO
> >> totally correct, triggers a warning - but I'm sure John can give
> >> us the correct interpretation of the standard C semantics).
Whatever the standard semantics are, a warning in this case is IMHO ok.
Given then warning the compiler could/should produce whatever the
standard requires.
But I can easily see, how a compiler could easer produce better code, if
it knew that NaN == NaN will be the same as 0 == 0. (E.g., just compare
two registers and not compile code to check vor a special NaN value.)