Re: gcc 2.95.3-1 and -mwin32

["Trevor Forbes" <trevorforbes at ozemail dot com dot au>]

----- Original Message -----
From: "Christopher Faylor" <cgf@redhat.com>
To: <cygwin-apps@cygwin.com>
Sent: Sunday, 18 March 2001 7:58
Subject: gcc 2.95.3-1 and -mwin32


> I've got gcc 2.95.3-1 ready to go.  It seems to work ok.
>
> I've had a private discussion with DJ about the affect of -mwin32.  We
> both feel that the windows header files should be included regardless of
> whether this switch is included or not.  I think that Chuck Wilson also
> persuasively argued this point.
>
> If we don't do this, then there are a lot of gratuitous changes required
> for a lot of configurations.  I've had to change four or five directories
> on sources.redhat.com alone.
>
> So, while I appreciate all of the work that Robert Collins is doing to get
> things working in autoconf, I've come to believe that the marginal
benefits
> of not including /usr/include/w32api outweigh the drawbacks.
>
> So, unless someone has a strong objection, the default for gcc 2.95.3-1,
> gcc 3.0, and gcc 3.1 will be to include /usr/include/w32api.
>
> I'll wait a couple of days to see if anyone has any strong objections
> before I release the new version of gcc.
>
> cgf

I also agree with /usr/include/w32api inclusion therefore I am more than
happy with this change.

However, I think w32api should be neutral with regards to WIN32 defines.

For example, I think the following should be "removed" and allow WIN32 to be
defined/not defined in gcc or (my preference) defined in the software being
ported:

w32api/windef.h

#ifndef WIN32
#define WIN32
#endif
#ifndef _WIN32
#define _WIN32
#endif

Comments?


Regards Trevor