Date:	Fri, 28 Oct 94 14:12:51 EDT
From:	peprbv AT cfa0 DOT harvard DOT edu (Bob Babcock)
To:	A DOT APPLEYARD AT fs1 DOT mt DOT umist DOT ac DOT uk
Cc:	djgpp AT sun DOT soe DOT clarkson DOT edu
Subject:	Re: C++ misc suggestions / queries
Reply-To:	babcock AT cfa DOT harvard DOT edu

>   (2) The Great Enemy of Gnu and of Gnu-users seems to be things loaded high.

Isn't the problem things that are _not_ loaded high?  GO32 requires ~130K of
memory in the first 640K, and recursive makes may require 3 (or more) copies
of GO32 to be loaded at the same time, tripling this requirement.  Loading
things high gets them out of the way.  Any attempt to temporarily move things
out of the way is likely to fail.  Those programs in the way are probably
TSRs and device drivers handling interrupts and you have to worry about how
to service those interrupts.

>   (3) When go32 finds that the PC is in V86 (or whatever is it) mode instead
> of in real mode, why can't it simply change the mode to real?

If the PC is in virtual 86 mode, it must have been put into that mode by some
software running at a higher privelege level.  The hardware won't let GO32
get out of V86 mode, and almost certainly something would break if you did
find a way out.

>   (4) How much work would it involve to give djgpp a mode that compiled and
> assembled and linked to make an .EXE file (not needing go32 to run it) in
> non-paged mode like Turbo C does? That is e.g. so that people developing Gnu
> C don't have to clutter their hard disks up with Turbo C as well as Gnu C

DJ has stated that one of the goals of version 2.0 will be to have a
stand-alone system which can be rebuilt without requiring a commercial
compiler.  Personally, I don't consider this very important.  Turbo C 2.0
doesn't take much disk space, and I think it's the best C compiler Borland
ever made, so I use it for production work anyway.  But politically it is
desirable to have builing GNU software not require commercial software.

- Raw text -