Re: [IDEA+RFC] Possible solution for min()/max() war

[Peter T. Breuer]

"A month of sundays ago David Woodhouse wrote:"
> 
> [e-mail] said:
> >  You got me curious enough to try it.  It compiles and links fine with
> > -O1 and higher under
> 
> >        gcc version 2.95.2 20000220 (Debian GNU/Linux)
> >        gcc version 2.8.1
> >        gcc version 2.7.2.3 
> 
> Oh well, then it _must_ be safe then - gcc has never changed unspecified 
> behaviour on us before, has it?

Well, I understand what you mean, but if the linux kernel wants it
and the C spec doesn't forbid it, then it'll either stay that way
or an "official" way will be found of evoking the desired behaviour.

The kernel already relies on -O1 expanding outb().

> The gcc engineer who took one look at the __buggy_udelay cruft, raised his
> eyebrows, swore and wandered off muttering must just have been having a bad
> day or something.

Actually, I was a bit more worried that 

  const unsigned int i = 1;
  if (i < 0)
      foo();

would generate warnings about the comparison always failing. But it
doesn't, at least not with -Wall. It does generate a warning about
implicitly declaring the function foo()! So I guess one has to add a
decl for it just above the call.  But it doesn't emit code for the call
itself, so the link is fine.

OTOH I now can't get #__LINE__ to expand as I want it where I want it.

Peter