Re: Flex: Cygwin and Windows ports

[Vincent Caron]

Bruce Lilly wrote:
> Vincent Caron wrote:
>
> > All right, so here _WIN32 should be replaced by _MSC_VER which is 
> > specific to the Visual compiler series.
>
> NO!!!  You have completely missed the point. The compiler is a
> separate issue from the compilation environment, the libraries,
> and the run-time environment.  The MS compiler will always
> define _WIN32 and _MSC_VER even when used in an environment
> which has unistd.h, htons, int64_t, etc.

I see. But the compiler and the environment are closely tied. It seems I 
am a rare and crazy beast trying to use Flex with Cygwin and MS tools 
(at least for the post-2.5.4 series). I find it pretty insane - or at 
least rather academic - trying to build Flex with Cygwin and a 
non-cygwin env, or with MSVC and a non-MSVC env.

Anyway, would you suggest to use a new macro to denote the MS default 
environment ? Like PLAIN_BORING_MS_WORLD to replace _WIN32 ? :)


> Use the mechanism provided by configure, e.g. for unistd,h
> use the HAVE_UNISTD_H macro.

Okay, so the patch in scan.l should read :

#ifndef HAVE_UNISTD_H
#define YY_NO_UNISTD_H
#endif

In order I can compile flex-2.5.23 with a scan.c generated by flex-2.5.22.