Subject: Re: lib/17630: atof(3) lacks of some precision
To: None <netbsd-bugs@netbsd.org>
From: Eric Jacoboni <jaco@teaser.fr>
List: netbsd-bugs
Date: 07/18/2002 13:05:56
>>>>> "David" == David Laight <david@l8s.co.uk> writes:

David> The only difference in netbsd between atof(x) and strtod(x,0)
David> is that the former uses slightly more stack!  I call that
David> equivalent.

I understand your point: from NetBSD point of vue (and Free/Open and
Linux for that matter), atof(x) is just a call to strtod(x, NULL). It
conforms to the standard as this document says only that atof()
doesn't need to set errno, it doesn't say that it cannot.

What makes me trouble in this manpage is that it seems to indicate
that atof() /is/ equivalent to strtod(), while this is only a (good)
NetBSD decision. From the standard point of vue, both functions are
not totally equivalent. I suppose some Unices (apart from those
mentionned above) may have choose another way, so it is a
implementation choice, isn't it?

That's why I think it would be clearer to add a IMPLEMENTATION NOTE or
add this note to the STANDARDS section.

-- 
Éric Jacoboni, né il y a 1330433663 secondes