Windows port (was Re: hey listen to this!!!!!)

[Stan Shebs <shebs at shebs dot cnchost dot com>]

"James R. Dunson" wrote:
> 
> At 10:17 PM 7/6/00 -0700, Stan Shebs wrote:
> >Upgrade your machine to Linux! :-)  But seriously, we don't yet
> >have a final release version of 7.3 built for Windows.
> 
>   How to put this carefully... There are machines that I use, or
> even manage, that I may run Xconq on, that I have no choice of
> OS due to other factors, and are running Windows.  Still other
> machines may run Windows due to the need to run specific apps
> already purchased for Windows and that do not have available
> free/cheap Linux equivalents.

It's very true.  You may remember from the post-7.2 days, when I was
looking at toolkits, that one of the requirements was that it had to
have a working Windows port, and that was one of the big selling points
for tcl/tk.

Indeed, with NT, things went pretty smoothly, once I worked around
the many unimplemented things in tk's Windows port.  But since I
left Cygnus, I no longer have access to NT systems, and W98 has
been a nightmare - not only does it crash a lot while building Xconq,
it crashes routinely while running it.  Is it Xconq? Is it tcl/tk?
Is it cygwin?  Or is it just W98 being its usual flaky self?

>   I've been thinking some lately about where to go with the modified
> discover.g I've been working with, but have put off any more work until
> I have a final release version to work with; it's difficult enough to
> figure out what does and doesn't work at any time to fight through
> it over and over again.  (I'm optimistically assuming that anything
> that works in a final release version will not stop working, at least
> for a while.)

Do you mean GDL, or Xconq behavior in general?  Intentional changes
are usually posted on this list, and unintentional changes should be
reported as bugs.  It's also important to report any bad consequences
of changes to this list; Keir for instance has pointed out some major
snafus in the code, and once we understood what was going on, the fix
was easy.  Any time you're having to fight with something, please don't
suffer in silence; tell the rest of us about it!

>   One of the things I have wondered is whether it is possible to
> compile Xconq with the free Borland C++ compiler.  The availability
> of easy and free DOS compiles of Angband and assorted related
> roguelikes via DJGPP has been very helpful in terms of increasing
> interest and available effort (more eyeballs, bigger bazaar, or
> whatever metaphor you prefer).
> 
> If a procedure and/or makefile could be developed for free Borland C++,
> it would greatly increase the amount of contribution that would be
> practical for those of us who spend much of our days stuck in a Windows
> world.

Hmmm, interesting though.  I take it the free version of cygwin is
problematic somehow?

In theory setting up for a new compiler shouldn't be too hard.  After
all, Xconq builds on Macs using CodeWarrior, which doesn't even
*have* makefiles, so it doesn't need any tricky build stuff.  tcl/tk
is likely to be more of a problem, since it seems to have been built
only with cygwin or VC++.

In any case, I'm not likely to be able to do this myself.  Now that
I'm working for Apple again, I only have access to two non-Macs: an old
machine that runs W95, and my new Dell laptop that runs Linux and W98,
with the W98 MTBF running around 1-2 hours - so guess which OS I tend
to stay in!  Nevertheless, I'm ready with advice and info for anybody
working on the Windows port, just ask.

Stan