14 Aug
2012
14 Aug
'12
11:45 p.m.
Is there anyone here who understands 16-bit DOS coding well enough to help
me make some 32-bit code work in 16-bit DOS?
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
15 Aug
15 Aug
12:52 a.m.
On Tue, 14 Aug 2012, David Griffith wrote:
Is there anyone here who understands 16-bit DOS coding
well enough to help me
make some 32-bit code work in 16-bit DOS?
Some background stuff: the problems seem to hinge on the fact that
unsigned longs are used all over the place. I'm thinking that if somehow
I can come up with a crutch for processing these variables, I could make
it work.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
1:08 a.m.
Are you running on a 16-bit machine (8086) or are you simply
running in DOS in real-mode on a 32-bit capable processor (80386+)?
If the latter, just get a DOS extender, like the one that comes with
DJGPP. Code written for 32-bit machines frequently makes
assumptions such as that sizeof(int) == sizeof(char*). With
an extender, you get a 32-bit address flat address space and
it will just work.
--Bill
--- On Wed, 8/15/12, David Griffith <dgriffi at cs.csubak.edu> wrote:
From: David Griffith <dgriffi at cs.csubak.edu>
Subject: Re: 32-bit code in 16 bits
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>
Date: Wednesday, August 15, 2012, 12:52 AM
On Tue, 14 Aug 2012, David Griffith
wrote:
Is there anyone here who understands 16-bit DOS
coding
well enough to help me make some 32-bit code work in 16-bit
DOS?
Some background stuff: the problems seem to hinge on the
fact that unsigned longs are used all over the place.?
I'm thinking that if somehow I can come up with a crutch for
processing these variables, I could make it work.
-- David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read
text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
6:11 a.m.
Is there anyone here who understands 16-bit DOS coding
well enough to help
me make some 32-bit code work in 16-bit DOS?
Is there some reason you can't use DPMI?
--
------------------------------------ personal: http://www.cameronkaiser.com/ --
Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com
-- Glory is fleeting, but obscurity is forever. -- Napoleon Bonaparte ---------
3:07 p.m.
On Wed, 15 Aug 2012, Cameron Kaiser wrote:
Yes. I'm trying to make the program runnable on 16-bit DOS machines
(8088, 80286). I might make a version for DPMI-capable machines, but this
is the last nasty chunk to crunch before I declare the program mature.
For the curious, it's Frotz at https://github.com/DavidGriffith/frotz
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
Is there
anyone here who understands 16-bit DOS coding well enough to help
me make some 32-bit code work in 16-bit DOS?
Is there some reason you can't use DPMI?
4:07 p.m.
--- On Wed, 8/15/12, David Griffith <dgriffi at cs.csubak.edu> wrote:
Then there's no magic bullet here. You either need a compiler that
implements double-precision integer arithmetic, e.g., synthesizing
32-bit arithmetic from multiple 16-bit operations, or rework the
program logic to operate with smaller integers.
--Bill
Is there some
reason you can't use DPMI?
Yes.? I'm trying to make the program runnable on 16-bit
DOS machines (8088, 80286).? I might make a version for
DPMI-capable machines, but this is the last nasty chunk to
crunch before I declare the program mature. 
8:47 a.m.
On 14 Aug 2012 at 23:45, David Griffith wrote:
Is there anyone here who understands 16-bit DOS coding well enough to
help me make some 32-bit code work in 16-bit DOS?
The solution depends on what you mean as "32 bit code". If you're
talking about code that uses the full set of 32-bit registers, but
otherwise stays within the bounds of 16-bit real mode addressing
(640K), it's simple--you simply tell the assembler that your program
is running in 16-bit real mode and use the 32-bit registers
explicitly. The proper code will be generated. (I can send you
samples if you'd like).
Otherwise, you're going to flip the operating mode of the CPU. Do
you want to run in segmented/protected mode or "flat 32 bit"mode
where 32-bit addressing across the entire memory space of the CPU is
used? In both cases, some sort of DPMI (DOS Protected Mode
Interface) server is required to handle allocation of memory and to
switch the CPU back to 16-bit real mode whenever a DOS system call or
BIOS call is made.
--Chuck
4:15 p.m.
On Wed, 15 Aug 2012, Chuck Guzis wrote:
On 14 Aug 2012 at 23:45, David Griffith wrote:
The basic problem appears to be that the code makes use of long ints.
DOS-16 mangles that, giving you only the bottom half.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
Is there anyone here who understands 16-bit DOS
coding well enough to
help me make some 32-bit code work in 16-bit DOS?
The solution depends on what you mean as "32 bit code". If you're
talking about code that uses the full set of 32-bit registers, but
otherwise stays within the bounds of 16-bit real mode addressing
(640K), it's simple--you simply tell the assembler that your program
is running in 16-bit real mode and use the 32-bit registers
explicitly. The proper code will be generated. (I can send you
samples if you'd like).
Otherwise, you're going to flip the operating mode of the CPU. Do
you want to run in segmented/protected mode or "flat 32 bit"mode
where 32-bit addressing across the entire memory space of the CPU is
used? In both cases, some sort of DPMI (DOS Protected Mode
Interface) server is required to handle allocation of memory and to
switch the CPU back to 16-bit real mode whenever a DOS system call or
BIOS call is made.
5:42 p.m.
On 15 Aug 2012 at 16:15, David Griffith wrote:
The basic problem appears to be that the code makes
use of long ints.
DOS-16 mangles that, giving you only the bottom half.
What are you using for a C compiler? Most 16-bit DOS C compilers and
runtimes handle "long" and "unsigned long" just fine as 32-bit
numbers.
If it's the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int type-
-define it as "int" for the large platforms and "long" for the 16-
bitters.
--Chuck
7:22 p.m.
On Wed, 15 Aug 2012, Chuck Guzis wrote:
On 15 Aug 2012 at 16:15, David Griffith wrote:
I'm using Turbo C++ 3.00. It's definitely chopping off half of the bits.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
The basic problem appears to be that the code
makes use of long ints.
DOS-16 mangles that, giving you only the bottom half.
What are you using for a C compiler? Most 16-bit DOS C compilers and
runtimes handle "long" and "unsigned long" just fine as 32-bit
numbers.
If it's the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int type-
-define it as "int" for the large platforms and "long" for the 16-
bitters. 
7:37 p.m.
I'm using Turbo C++ 3.00. It's definitely
chopping off half of the
bits.
Is the code anything you can let people like us look at? I'd be
willing to have a look. It seems to me it could be a bug in the code -
it's easy to truncate things to int without realizing it - and I'm
moderately good at spotting such things.
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
8:51 p.m.
On Wed, 15 Aug 2012, Mouse wrote:
Please look. It's at https://github.com/DavidGriffith/frotz. If you need
a ready-to-use development environment, let me know. The problem code is
in the blorb/ subdirectory.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
I'm using
Turbo C++ 3.00. It's definitely chopping off half of the
bits.
Is the code anything you can let people like us look at? I'd be
willing to have a look. It seems to me it could be a bug in the code -
it's easy to truncate things to int without realizing it - and I'm
moderately good at spotting such things.
7:37 p.m.
On 15 Aug 2012 at 19:22, David Griffith wrote:
I *knew* there was a reason that I stayed away from Borland.
Microsoft C handles 32-bit ints just fine. Are you certain it's not
just a matter of a misplaced or missing cast?
As long as you don't mind restricting the thing to 386+ platforms,
you can use something like DJGPP that comes with its own DPMI server.
--Chuck
If it's
the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int
type- -define it as "int" for the large platforms and "long" for the
16- bitters.
I'm using Turbo C++ 3.00. It's definitely chopping off half of the
bits. 
8:17 p.m.
On 8/15/2012 7:37 PM, Chuck Guzis wrote:
On 15 Aug 2012 at 19:22, David Griffith wrote:
I *knew* there was a reason that I stayed away from Borland.
Microsoft C handles 32-bit ints just fine. Are you certain it's not
just a matter of a misplaced or missing cast?
As long as you don't mind restricting the thing to 386+ platforms,
you can use something like DJGPP that comes with its own DPMI server.
--Chuck
i have a microdata 32/s simulator written in borland c. A friend wrote
it before he passed away, and I don't have good access to his toolchain,
to you got what you got. It's been a real chore trying to get it to
work in gcc.
We discussed using gcc before he stated coding and he had borland on his
home computer. maybe someday I'll get it to work with gcc, but for now
he took maximum advantage of the 16 bit'ness since there were a lot of
16 bit operations the hardware did very well.
If
it's the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int
type- -define it as "int" for the large platforms and "long" for the
16- bitters.
I'm using Turbo C++ 3.00. It's definitely chopping off half
of the
bits.
8:54 p.m.
On Wed, 15 Aug 2012, Chuck Guzis wrote:
On 15 Aug 2012 at 19:22, David Griffith wrote:
I *knew* there was a reason that I stayed away from Borland.
Microsoft C handles 32-bit ints just fine. Are you certain it's not
just a matter of a misplaced or missing cast?
As long as you don't mind restricting the thing to 386+ platforms,
you can use something like DJGPP that comes with its own DPMI server.
I'm quite sure the casts are okay. So, if I somehow dig up an old
Microsoft C compiler, I should be fine? I'm trying to keep this thing
compatible with pre-386 for the simple reason of playing old Infocom games
on old machines.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
If
it's the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int
type- -define it as "int" for the large platforms and "long" for the
16- bitters.
I'm using Turbo C++ 3.00. It's definitely chopping off half of the
bits.
9:20 p.m.
On 15 Aug 2012 at 20:54, David Griffith wrote:
I'm quite sure the casts are okay. So, if I
somehow dig up an old
Microsoft C compiler, I should be fine? I'm trying to keep this thing
compatible with pre-386 for the simple reason of playing old Infocom
games on old machines.
Should be--I've got tons of old MSVC 16-bit code that uses 32 bits.
Some of it even uses 64 bit integers (a little assembly library
there). No problems at all.
MSVC 8.00c is very stable. I think it was the last 16-bit C that
Microsoft shipped.
--Chuck
16 Aug
16 Aug
8:06 a.m.
On 8/15/12 10:54 PM, David Griffith wrote:
I'm quite sure the casts are okay. So, if I
somehow dig up an old
Microsoft C compiler, I should be fine? I'm trying to keep this thing
compatible with pre-386 for the simple reason of playing old Infocom
games on old machines.
I have a box of MS C v5.0. IIRC it's on CD, not floppy, and comes
with docs.
Want it?
Doc
8:29 a.m.
On Thu, 16 Aug 2012, Chuck Guzis wrote:
On 16 Aug 2012 at 10:06, Doc wrote:
It's listed as Visual C++ 1.52.
I'd still recommend Open Watcom - it wipes the floor with just about every
other DOS based C compiler out there. It'll also generate Win16/32
binaries and OS/2 if memory serves.
g.
--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.
ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!
Buying desktop hardware and installing a server OS doesn't make a
server-class system any more than sitting in a puddle makes you a duck.
[Cipher in a.s.r]
I have a box of MS C v5.0. IIRC it's on
CD, not floppy, and comes
with docs.
The last (1.52c/8.00c?) version of 16-bit MSVC is available free for
download if you know a MSDN subscriber.
8:40 a.m.
On Thu, Aug 16, 2012 at 08:29:18AM -0700, geneb wrote:
On Thu, 16 Aug 2012, Chuck Guzis wrote:
There is also Digital Mars which ought to be pretty good.
--
_ _
._ _ _ <_> ___ _ _ ___ ___ ___ _| | ___
| ' ' || |/ | '| '_>/ . \/ | '/ . \/ . |/ ._>
|_|_|_||_|\_|_.|_| \___/\_|_.\___/\___|\___.
On 16 Aug 2012 at 10:06, Doc wrote:
It's listed as Visual C++ 1.52.
I'd still recommend Open Watcom - it wipes the floor with just about
every other DOS based C compiler out there. It'll also generate
Win16/32 binaries and OS/2 if memory serves. I have a box of MS C v5.0. IIRC it's on
CD, not floppy, and comes
with docs.
The last (1.52c/8.00c?) version of 16-bit MSVC is available free for
download if you know a MSDN subscriber.
8:54 a.m.
On 16 Aug 2012 at 8:29, geneb wrote:
It's listed as Visual C++ 1.52.
Which is an 8.00c compiler (Microsoft's strange idea of calling the
whole IDE platform by a different version than the compiler itself.
Go figure.)
It's a DOS compiler, only sort-of. It won't run on anything less
than an 80386, if memory serves.
I'd still recommend Open Watcom - it wipes the
floor with just about
every other DOS based C compiler out there. It'll also generate
Win16/32 binaries and OS/2 if memory serves.
That's certainly another option. It always seemed to me that MSVC
had more aggressive optimization, however. (/Ox switch).
--Chuck
9:19 a.m.
On Thu, 16 Aug 2012, Chuck Guzis wrote:
On 16 Aug 2012 at 8:29, geneb wrote:
Watcom C was always the choice of game developers that wanted really good
performance. It's my understanding that the Watcom optimizer was pretty
much unparalleled until the last decade or so.
g.
--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.
ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!
Buying desktop hardware and installing a server OS doesn't make a
server-class system any more than sitting in a puddle makes you a duck.
[Cipher in a.s.r]
It's listed as Visual C++ 1.52.
Which is an 8.00c compiler (Microsoft's strange idea of calling the
whole IDE platform by a different version than the compiler itself.
Go figure.)
It's a DOS compiler, only sort-of. It won't run on anything less
than an 80386, if memory serves.
I'd still recommend Open Watcom - it wipes
the floor with just about
every other DOS based C compiler out there. It'll also generate
Win16/32 binaries and OS/2 if memory serves.
That's certainly another option. It always seemed to me that MSVC
had more aggressive optimization, however. (/Ox switch).
1 p.m.
On 16 Aug 2012 at 9:19, geneb wrote:
Watcom C was always the choice of game developers that
wanted really
good performance. It's my understanding that the Watcom optimizer was
pretty much unparalleled until the last decade or so.
It could be a matter of to-may-toe vs. to-mah-toe.
Around 1994, I was really in the market for a good 32-bit C, so I
bought every compiler I could find for Windows/DOS. I was appalled
at Watcom's poor optimization in 32-bit mode and put the whole thing
aside in favor of, eventually, Microsoft (which I'd been using since
it was Lattice C.
My test consisted of a few small programs that tested common
optimizations, such as common subexpression elimination, removal of
invariant expressions from loops, reduction in strength, dead
variable removal, bottom-of-loop prefetch, and so on. I examined
the generated code, but did not run any timed benchmarks.
About two years ago, I gave away the almost-unused boxed set of
Watcom C that I'd purchased. I saw no reason to hang onto it.
FWIW,
Chuck
4:13 p.m.
On Thu, 16 Aug 2012, geneb wrote:
On Thu, 16 Aug 2012, Chuck Guzis wrote:
I was playing around with Open Watcom last night, but couldn't get my head
wrapped around the weird way it expects Makefiles to be written. Turbo
C++ Makefiles are essentially plain old Posix.
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
On 16 Aug 2012 at 10:06, Doc wrote:
It's listed as Visual C++ 1.52.
I'd still recommend Open Watcom - it wipes the floor with just about every
other DOS based C compiler out there. It'll also generate Win16/32 binaries
and OS/2 if memory serves. I have a box of MS C v5.0. IIRC it's on
CD, not floppy, and comes
with docs.
The last (1.52c/8.00c?) version of 16-bit MSVC is available free for
download if you know a MSDN subscriber.
17 Aug
17 Aug
5:34 a.m.
On Thu, 16 Aug 2012, David Griffith wrote:
On Thu, 16 Aug 2012, geneb wrote:
Why not just grab the GNU make (there has to be a DOS port around
somewhere) and use that? Even Borland's make should do the job, just call
Watcom instead.
g.
--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.
ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!
Buying desktop hardware and installing a server OS doesn't make a
server-class system any more than sitting in a puddle makes you a duck.
[Cipher in a.s.r]
On Thu, 16 Aug 2012, Chuck Guzis wrote:
I was playing around with Open Watcom last night, but couldn't get my head
wrapped around the weird way it expects Makefiles to be written. Turbo C++
Makefiles are essentially plain old Posix. On 16 Aug 2012 at 10:06, Doc wrote:
It's listed as Visual C++ 1.52.
I'd still recommend Open Watcom - it wipes the floor with just about every
other DOS based C compiler out there. It'll also generate Win16/32
binaries and OS/2 if memory serves. I have a box of MS C v5.0. IIRC it's on
CD, not floppy, and comes
with docs.
The last (1.52c/8.00c?) version of 16-bit MSVC is available free for
download if you know a MSDN subscriber.
16 Aug
16 Aug
6:28 a.m.
On Wed, 15 Aug 2012, David Griffith wrote:
On Wed, 15 Aug 2012, Chuck Guzis wrote:
Dave, I'd give Open Watcom a shot - it includes a DOS extender if you need
it.
g.
--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.
ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!
Buying desktop hardware and installing a server OS doesn't make a
server-class system any more than sitting in a puddle makes you a duck.
[Cipher in a.s.r]
On 15 Aug 2012 at 16:15, David Griffith wrote:
I'm using Turbo C++ 3.00. It's definitely chopping off half of the bits.
The basic problem appears to be that the code
makes use of long ints.
DOS-16 mangles that, giving you only the bottom half.
What are you using for a C compiler? Most 16-bit DOS C compilers and
runtimes handle "long" and "unsigned long" just fine as 32-bit
numbers.
If it's the matter of the source (that 32-bit platforms treat "int"
as 32 bits), consider creating a conditional typedef for the int type-
-define it as "int" for the large platforms and "long" for the 16-
bitters.
4510
days inactive
4512
days old
25 comments
9 participants
participants (9)
-
cclist＠sydex.com -
dgriffi＠cs.csubak.edu -
doc＠vaxen.net -
geneb＠deltasoft.com -
jws＠jwsss.com -
microcode＠zoho.com -
mouse＠Rodents-Montreal.ORG -
spectre＠floodgap.com -
wmaddox＠pacbell.net