[PATCH] Document arithmetic overflow semantics

[Neil Booth]

Mark Hahn wrote:-

> yes!  I'm a lowly gcc end-user, but I'm apalled that gcc-gods would 
> even consider compromising optimization in favor of some nebulous 
> make-bugs-safer argument.
> 
> gcc is not a security-fix tool.  please permit users to select 
> absolute-max-optimization somehow; this is orthogonal to gcc's 
> extremely valuable diagnostics about undefined/questionable code.

Indeed.

One day I'd like to see GCC capable of much of what the Linux "Checker"
does, though.

Neil.