More c99status.html updates

[Zack Weinberg]

On Thu, Jul 13, 2000 at 07:28:56AM +0100, Joseph S. Myers wrote:
> On Wed, 12 Jul 2000, Zack Weinberg wrote:
> 
> > For features listed as "Broken", it'd be nice to put some explanation
> > somewhere - perhaps at the bottom, describe each brokenness in detail.
> 
> I'm inclined to think that some of the Missing entries should have some
> sort of explanation of what the missing feature is as well.

Agreed.

> > %lf in printf				Done
> 
> GCC doesn't yet recognise %lf.  Nor %t or %j (ptrdiff_t and intmax_t), nor
> %F (like %f but uppercases nan and inf), nor printf %zn, nor proper
> pedwarns for the new strftime formats in C89 mode, and it unconditionally
> follows the glibc kludge for scanf %as.

I stand corrected - I thought your patches had already been applied,
or something like that.  [What's scanf %as?]

> > [3] I had to look up "IEC 60559".  You should perhaps put 
> > (IEEE 751 floating point) after it.  Currently GCC implements this
> > correctly if and only if the hardware is perfectly compliant.
> 
> GCC also needs to define __STDC_IEC_559__ when conforming.  (This also
> seems to be tied up with the standard pragmas.)

And, unfortunately, glibc 2.1 unconditionally defines __STDC_IEC_559__
on every supported architecture.  [And some people think C99's use of
IEC 559 is utterly broken as designed, but that's another kettle of
fish entirely.  Do you read comp.std.c?  You should.]

There's a whole area of issues coming from interactions with the C
library, which the standard doesn't even contemplate and we
historically don't handle well.  Just look at <limits.h>, but be
prepared to lose sanity points.

zw