Re: gcc 2.95.3-1 and -mwin32

["Robert Collins" <robert dot collins at itdomain dot com dot au>]

----- Original Message -----
From: "Christopher Faylor" <cgf@redhat.com>
To: <cygwin-apps@cygwin.com>
Sent: Sunday, March 18, 2001 1:58 PM
Subject: Re: gcc 2.95.3-1 and -mwin32


> On Sun, Mar 18, 2001 at 01:05:34PM +1100, Robert Collins wrote:
> >> w32api/windef.h
> >>
> >> #ifndef WIN32
> >> #define WIN32
> >> #endif
> >> #ifndef _WIN32
> >> #define _WIN32
> >> #endif
> >>
> >> Comments?
> >
> >I don't think the software being ported should ever define WIN32
itself.
> >It'd be like defining glibc_version (or whatever that value is
called)
> >yourself.
> >
> >Setting it for gcc is one thing, and quite appropriate. As far as
w32api
> >goes, I haven't investigated and thus won't comment..
>
> Including /usr/include/w32api and not allowing the header files to
define
> WIN32 constants would be equivalent to not including
/usr/include/w32api
> at all.  The users would get confusing error messages.
>
> Microsoft's windows headers seem to define WIN32 constants in a number
of
> places.  I don't see anything wrong with this practice.
>
> cgf
>

I don't see anything wrong with w32api defining WIN32 either - I was
arguing against changing the current practice. Probably worded badly
though..

Rob