Sender:	<solyom AT bmeik DOT eik DOT bme DOT hu>
Date:	Tue, 07 Sep 1993 10:54:09 +0100
From:	solyom AT bmeik DOT eik DOT bme DOT hu
Reply-To:	SOLYOM AT HUBME51 DOT bitnet
To:	DJGPP AT SUN DOT SOE DOT CLARKSON DOT EDU
Subject:	DPMI 0.9 is good enough to DJGPP!

I have read a lot about that I cannot run GCC or GCC compiled programs under
MS Windows, because it only supports DPMI 0.9 and DJGPP needs 1.0. As I have
found out this piece of information is not correct. I am using QEMM V. 7.0
as my memory manager. This version gives DPMI services V0.9 with 32 bit support.
And it works well with DJGPP! So it seems that the problem with MS windows is
that it does not support 32 bits. Although you can say that DPMI 0.9 with
32 bit support is DPMI 1.0, but I don't think so. If it was, then Quarterdeck
would claim that they are the first to support DPMI 1.0. This is of course
only an indirect proof, but I really do not know much about DPMI...

Please forgive me if in this letter I pose some qustions too concerning DJGPP!
1. I cannot use Go32 version 2.2.2 because it complains for a file named
2.2.2 which is an invalid name for a DOS file. What is the problem?
2. I wanted to use system() calls from my protected program. Most of the
time I received 'Segmentation fault messages in different parts of my
program. I tried to debug it, and I think it is a stack problem. And not only
this! EVen if I does not receive the segmentation fault message, an open
file is left at my hard disk, which results in lost clusters (from 1 to
several megabytes)
3. I patched Go32 to deal with the extended Open function (0x6c). I could only
compile a working version with Turbo C 1.5. I compiled it with Borland C++ 3.1
too but the compiled go32.exe didn't work! Why? And I could not create a working
version of the debugger.

Thanks for any help!

- Raw text -