Newsgroups: rec.games.int-fiction
Path: gmd.de!xlink.net!howland.reston.ans.net!spool.mu.edu!olivea!pagesat!news.cerf.net!kaiwan.com!cinnamon
From: cinnamon@kaiwan.com (R. Dominick)
Subject: Re: help with inform!
Message-ID: <CFpJnD.G0q@kaiwan.com>
Organization: Coffeehouse Software, Inc.
X-Newsreader: TIN [version 1.2 PL2]
References: <2ar2md$eos@news.u.washington.edu>
Date: Sat, 30 Oct 1993 11:23:37 GMT
Lines: 30

spoof@u.washington.edu wrote:

> arrrrgggh!!!  can someone help me with compiling inform?!
> i have the MSC7.0 compiler, and i keep getting an error about
> having too many communal variables.  i have NO IDEA how to 
> fix this.   anyone?

Under MS-DOS, *if* you go up to the Huge memory model, you can only
have 64K of global variables in one module of your program.  All
through the program, somewhere around 90K of globals are defined, and
that's in one module.  Begin to see the problem?

Earlier today, I broke Inform all up into several different files, and
made headers for it, and so on; I got an executable, but one with the
same problems both I and another programmer had before: being unable
to compile very short programs (i.e. the "Hello, Cruel World" thing)
and having very strange problems with the "Deja Vu" example.  I've got
a weekend coming up, and, unless I get called for an emergency
gotta-do-it-right-now temporary assignment, I was planning on building
a debug version of the code and seeing exactly what goes on.

By changing the call to strtoul and typecasting the return result of
strtol, I was able to get Inform to compile with GCC under SunOS 4;
thid's probably work with DJGPP, but since I don't currently have the
iron to run it...

Ah, well.  At least Inform is in 9 small pieces now, so one change
doesn't take seven years to recompile and link.
-- 
i dreamt that one had died in a strange place
