Re: make -mno-win32 the gcc default?

["Larry Hall (RFK Partners, Inc)" <lhall at rfk dot com>]

At 05:00 PM 1/10/2001, Christopher Faylor wrote:
>On Wed, Jan 10, 2001 at 04:36:41PM -0500, Larry Hall (RFK Partners, Inc) wrote:
> >At 04:31 PM 1/10/2001, Christopher Faylor wrote:
> >>So, what does everyone think about making -mno-win32 the default for Cygwin?
> >>
> >>This would mean that WIN32 variables would not be defined by gcc by default.
> >>It would require a -mwin32 switch for these to be activated.
> >>
> >>Earnie has been proposing this and I sort of like the idea but dread the
> >>potential hue and cry that could result.
> >>
> >>So, is the hue and cry from the lack of WIN32 counterbalanced by potentially
> >>easier UNIX porting?
> >
> >
> >Yes.  It fits with the stated goals of Cygwin.  We can counter the outcry 
> >with a nice FAQ so we at least will not have to repeat ourselves constantly.
> >
> >I'm in!
>
>You mean like this?
>
>     Why am I getting errors from "windef.h"?
>     Why am I getting errors saying simple things like "HANDLE" are undefined?
>     Why can't I compile my Windows program with Cygwin 1.1.8?  It worked fine
>     with B19!
>
>     Please send email to lhall@rfk.com and he'll explain everything.
>
>:-)
>
>cgf


Yes, something like that.;-)

Seriously though, I'm betting on the fact that its easier to answer questions
about why Win32 programs don't compile by default than it is to explain why
Cygwin won't build and run GNU/UNIX packages.  Of course, as Chuck Wilson
points out, there will be those packages which are ported to both UNIX and
Windows that seem to work right now because they make use of some Win32 
stuff that won't be allowed with this change.  While I expect most of these
will still build, they might break when running.  Questions about these 
will potentially be prolific at the beginning but at least the solution is
cleanly stated (port it properly) unless the functionality in Cygwin is 
missing.  There's some grey area but I expect the result for any strict 
Win32 program will be that it won't compile which is handled by 1 simple
FAQ (i.e. add the -mwin32 flag for Windows programs).  The result for any
UNIX port will be either the package won't compile or program errors/bugs 
(due to bad porting or lack of Cygwin support).  Another general FAQ can
handle that too.  While the initial volume of all these complaints may be 
high, it will tend to force the packages and Cywgin to get better.  Add to 
this the fact that we won't have people wondering why their favorite GNU 
package uses Win32 block of code under Cygwin instead of the UNIX/POSIX 
code and I think this will be a plus (or at least easier to explain!;-))  

Thus ends my diatribe on how things work in a perfect world.;-)



Larry Hall                              lhall@rfk.com
RFK Partners, Inc.                      http://www.rfk.com
118 Washington Street                   (508) 893-9779 - RFK Office
Holliston, MA 01746                     (508) 893-9889 - FAX



--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple