Re: working with addresses
From: Jumbo (nospam)
Date: Sat, 17 Jan 2004 10:53:08 -0000
"Micah Cowan" <firstname.lastname@example.org> wrote in message
> "Jumbo" <pcr1000011<nospam>@uko2.co.uk> writes:
> > "Micah Cowan" <email@example.com> wrote in message
> > news:firstname.lastname@example.org...
> > > Ian Woods <newspub2@wuggyNOCAPS.org> writes:
> > >
> > > <snip>
> > >
> > > > > Conclusion: either gcc is hopelessly broken, or an object file is
> > > > > a translation unit.
> > > >
> > > > Or, you didn't invoke the C compiler. :)
> > >
> > > No: he invoked the C implementation, which is all that was
> > > required. Also, note that the vast array of (required)
> > > diagnostics proves in itself that the object file is not a
> > > (valid) translation unit: though I suppose for the purposes of
> > > that single run it /was/ briefly a translation unit which invoked
> > > UB up the wazoo. But that's true of anything you should happen to
> > > run through a C implementation.
> > Rubbish.
> Not according to the C standard, without which the very term
> "translation unit" is utterly meaningless.
Was the file a C programming language file? No!
Are the C standards relevent in this case? No!
Do the C standards state that your implementation must only use C files? No!
Could an implementation take a java file? Yes.
Would the C standards be relevant? As with object files, No!
Just because the standards mention Linking as a Phase Of Translation from
C's point of view does not mean that the C standards are applicable every
time you invoke the linker.