Re: gcc 2.95.3-1 and -mwin32

["Charles S. Wilson" <cwilson at ece dot gatech dot edu>]

> > Charles Wilson wrote:
> > >Q: Does that also mean that the default for gcc-2.95.3-1 
> > >on cygwin will be to #define WIN32 & WINDOWS in addition 
> > >to merely including /usr/include/w32api in the include 
> > >path?  e.g. that -mwin32 is the default now (again)?

> Chris Faylor said:
> > I wasn't planning on doing this, no.  I still think that 
> > Cygwin should not define these unless asked.
> >
> > This does cause problems for some of the stuff in 
> > sources.redhat.com but they are minor compared to 
> > the effort of adding -mwin32 everywhere that
> > it was needed.
> >
> > So, this is still different from gcc-2.95.2-1.

Robert Collins asked:
> Will -mwin32 still be present? I think it should if you 
> do not define WIN32.

This complicates things somewhat; there would be *three* 
separate behaviors:

default: include /usr/include/mingw, but don't #define WIN32
  and friends.

-mwin32 (& -mno-cygwin): include /usr/include/mingw, and 
  #define WIN32 and friends.

-mno-win32: don't include /usr/include/mingw, and don't 
  #define WIN32.

BUT, these three behaviors make sense to me, given the 
experiences we've had with gcc-2.95.2-7 ... -9.

--Chuck