Re: [boost] Formal Review: any

From: Ed Brey (brey_at_[hidden])
Date: 2000-11-08 11:53:50

From: "Kevlin Henney" <kevlin_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Wednesday, 08 November 2000 8:30
Subject: Re: [boost] Formal Review: any

> In message <00ef01c0498e$eb894740$120524d4_at_pdimov>, Peter Dimov
> <pdimov_at_[hidden]> writes
> >From: "Kevlin Henney" <kevlin_at_[hidden]>
> >
> >> >A question, why the <cstdlib>? Note that MSVC's <c*> headers don't put
> >> >identifiers in std.
> >>
> >> Because any doesn't just compile with MSVC.
> >
> >Yes, of course, my point is that test-any.cpp doesn't use anything from
> ><cstdlib>, if I'm not mistaken; and if it did, you'd either get an error
on
> >MSVC that the identifier is not a member of std:: or an error from the
real
> >compilers that the identifier is in std::. <stdlib.h> would work on
both -
> >had it been required.
>
> EXIT_SUCCESS and EXIT_FAILURE, which are both gtee'd to be macros, are
> both used in test-any.cpp.

Just to clarify, even if a non-macro identifier from <cstdlib> were used,
including <stdlib.h> is not the preferred workaround to the VC/Dinkum bug
regarding namespace std.  We have already hashed this issue a couple times.
You can find the workaround we came up with at the end of config.hpp in the
BOOST_NO_STDC_NAMESPACE section.