Re: [Mingw-users] Re: _WIN32?

[Kai Ruottu <karuottu@freenet.hut.fi>]

Christopher Faylor wrote:

> On Tue, May 01, 2001 at 09:19:39AM +1000, Danny Smith wrote:
> >
> >--- Christopher Faylor <cgf@redhat.com> wrote: > On Mon, Apr 30, 2001 at
> >01:05:17PM -0400, Andrew Cagney wrote:
> >> >Christopher Faylor wrote:
> >> >
> >> >>         _WIN32          Pure WIN32 support 'gcc -mno-cygwin' --
> >> >>                         don't know if it even builds.  _WIN32 is
> >> >>                         now no longer defined by gcc.
> >> >
> >> >So, can I delete this one? :-)
> >>
> >> I don't know.  Someone used to maintain a non-cygwin Windows version of
> >> gdb.  I don't know if it even builds, like I said.  I'm not even sure
> >> how to configure it.

 Perhaps this 'MSC_VER'-one, as the more detailed #ifdef's define it,
may be
removed. For me the _WIN32 has always been unclear.  It is more clear to
use
the '__CYGWIN32__' and '__MINGW32__' than the 'combined' _WIN32 meaning
them
both (at least earlier).

 If a port, another than for GCC, would be needed, let it at least be
for the
free or very cheap (coming with books and magazines) Borland
compilers(5.5 and
CodeBuilders), not for the very expensive MS-ones....

> >The last non-cygwin windows version of gdb that I know of is 4.18 which Mumit
> >made available here:
> > ftp://ftp.xraylith.wisc.edu/pub/khan/gnu-win32/mingw32/ports
> >
> >The diff file for that port is dated May 1999.
> >
> >I have just started work on gdb 5.0 for mingw. I have not gotten far, mainly
> >because of lack of time.  Has anyone else made progress.

 Yes, with both 4.18.x and 5.0  using the Insight-derivation. The
snapshots
weren't yet looked by me but I wouldn't expect big surprises. In order
to get
something to download, I put the latest incarnations of these (the
binaries,
DLLs and scripts, some of these a year or more old though) for download
via
my new homepage :

     http://koti.mbnet.fi/"karuottu

 Perhaps some day I will try to collect the diffs against something (may
be
hard when adapting stuff when needed from the snapshots has been my
habit...)

> If you are at least considering this, then it sounds like the _WIN32 conditionals
> in gdb sources should not be removed.
>
> I assume that they will be required if you build a no-cygwin version of gdb,
> correct?

 As said, they just make the things more unclear. Questions like "Does
this work
under Cygwin, Mingw, MSVC++, BorlandC++, WatcomC,... ?" do arise,
because the
_WIN32 seems to mean them all. If only the Cygwin and Mingw ports are
only used
to compile GDB/Insight, why not use only the '__CYGWIN32__' and
'__MINGW32__' ?

 Removing the special Cygwin '_imp____cygwin_environ' etc. stuff, still
in the
snapshots (e.g. 'tcl/generic/tclEnv.c') when I last saw them, should be
sane...
These things should be defined in the Cygwin standard headers, not in
the GDB/tcl/tk/tix/itcl etc. sources. I'm quite sure I stripped all
these kind of
stuff for Mingw away, it already being defined in the standard
headers...

Cheers, Kai