Mail Archives: djgpp-workers/2003/03/17/06:01:58
Hello.
Martin Stromberg wrote:
[snip]
> Richard said:
[snip]
> > Also, we seem to understand nan(<anything>), but ignore it.
>
> Indeed. I don't really mean to implement it. Somebody else will need
> to do that.
>
> Perhaps we should only accept "nan()" until somebody does implement
> it.
What I was trying to point out is that saying that we understand
nan(<anything>) in What's Changed might give the impression that we will do
something with <anything>. Maybe a note could be added to the strto* pages
about inf*, nan*? This could include a statement about <anything> being
ignored.
I think ignoring the <anything> in nan(<anything>) is fine.
> > > + /* If we are going to support "nan(0x1234) for setting specific bits,
> > > + * that code goes here. Something like "bits = strtoul( &s[4], &end_p,
> > > + * 0);".
> > > + */
>
> > Maybe you should mark this (and similar) comments with a TODO or
> > FIXME? FIXME seems to be the preferred to-do marker in the sources.
>
> Well. The support of what is within the parentheses are up to the
> implementation. So there's nothing wrong with it (nothing to FIX) as
> it is today. But I suppose a "FIXME, any hackers with twiching fingers
> out there? We would like to:" or something might be a good idea.
Yes.
> Whatever we decide to support should probably go into the
> documentation.
Yes.
> > Maybe we could add a portability note
> > saying that support for Inf, NaN, NaN() is C99-specific?
>
> Errmh... How would such a portability note look like? (I though those
> were c99, !c89 etc.)
@port-note <standard> <comments>
find src/libc -name '*.txh' | xargs grep port-note
Take a look at src/libc/ansi/stdio/scanf.txh. That has a C99-specific
portability note.
Bye, Rich =]
--
Richard Dawe [ http://www.phekda.freeserve.co.uk/richdawe/ ]
- Raw text -