ES>LR> You mean your personally owned consulting co. or the company
ES>LR> you are an employee of...

ES>Mine is an independentant consulting firm which specializes in custom
ES>scientific and engineering software and crypto applications. These are
ES>vertical, one-of-a-kind projects for large corporate and government entities

ES>LR> Why did you feel *compelled* to go to native OS/2 apps
ES>LR> ... if they [as DOS apps] ran well under OS/2's VDM?

ES>I don't like the word "compel" when applied to me, but in
this case it almos
ES>fits.  The prime reason is memory.  Some data sets for certain of my
ES>simulations require 100MB or more of memory; doing this with DOS is a real
ES>PITA.  With OS/2, 'tis a joy in comparison. A multi-threaded OS/2 app crunch
ES>lots of numbers makes DOS/WIN look like horse-and-buggy (no pun intended)
ES>technology.

Due to the kind of applications you implement I can see why you would
need not only the memory and multi-tasking advantages, but the precision
advantages as well to a 32-bit implementation(vs a 16-bit
implementation).

ES>Running DOS-based apps under UNIX is about as inane as doing so under OS/2 o
ES>NT.  It might work, but you're still stuck with all of DOS's limitations.  B
ES>by using C/C++ I am able to develop for OS/2 on Intel boxes and have been ab
ES>to expand to UNIX on workstations.

Not all applications need the same kind of exacting resources your
implementations need.  Inane for your purposes, I agree, but not inane
in general - nor for my purposes.

My apps are designed for ordinary individual consummers in high
volume and general appeal - not for businesses at present.  Hence, my
potential customers are not as technically demanding as yours seem to be
- technical in the mathematical and/or engineering sense.

My apps is so unique and it's appeal is so universal (IMO), that I
believe consummers will buy it in any form in which it is available that
they can get their hands on - DOS, Windows, OS/2, etc.  16-bit precison
is fine for my application, 32-bits is not necessary technically
speaking - from a mathematical perspective.

The appeal of OS/2 for my apps is it's market - those individual
consummers who prefer and need multi-tasking and like PM.  My rough
guess right now is that there aren't too many individual consummers
using OS/2 due to it's inherent operation complexities in comarison to
DOS.  Hence, it seems like only Pros, mostly, have already moved to
OS/2.

I recently heard that there are 5 million OS/2 users out there, and
I tend to believe that the majority of them are being used by Pros right
now. There is allegedly 50 million versions of Windows out there in
comparison.

What I don't know right now is what percentage of those 50 million
Windows are actually being used by individual consummers - or by
businesses for that matter.

OSTEAMS members seem to believe that most of those Windows versions
that came with their new computers are NOT being used in any significant
numbers. I remain less sure this is a fact.

DOS and Windows emulation under OS/2, and if it were, or is, available
under UNIX(as it may be - LINUX???, COHERENT???), allows people and
businesses to *gradually* move to 32-bit apps - especially when no
32-bit apps equal to their existing 16-bit apps exist.  A SOP for our
industry as originated by IBM for it's Mainframes.

100% upward compatibility being the crucial process that permits
selling newer computers with newer operating systems with profitable
add-ons.

Without such necessary emulation smoothing the way with 100 % upward
compatibility, a new operating system would come out and likely
no one would move to it if users would demand 32-bit apps
*immediately* -  and none existed.

ES>LR> ... whatever did Zak say to you in mid 1993 that convinced you
ES>LR> so thoroughly that you decided to move to C++ for OS/2?

ES>I think it was mostly his condescending attitude.  He sounded like the clown
ES>the White House, telling me to trust Borland (and by extension, Microsoft) -
ES>*they* knew what was good for me, and I didn't.  Factually, he stated that
ES>Borland was working on a 32-bit Pascal compiler that would target Windows
ES>(including NT, Chicago, and Cairo), although he would give no details.  He
ES>further said that Borland would continue to "watch" the
OS/2 market for Pasc
ES> The tone of his voice was enough to tell me what I wanted to know.

Yes, I can see that the Zak's condensending tone could have
turned anyone off and implied the conclusion you drew about Borland
NOT coming out with a 32-bit Pascal.  Still, in my view, it seems a
business form of insanity to abandon the alleged 3 million users of
Borland's Pascal compilers worldwide (Borland's claim) - where some 40
percent are allegedly Pros.

Fortunately, however, I recently heard a German business has released
a 32-bit Pascal Shareware compiler in beta test for OS/2(spos??),
earlier this summer, and allegedly they will come out with a full GA by
this September, 1994. In addition someone else told me Microway
(American Co., MA) has also recently released a 32-bit Pascal compiler
for OS/2.

Both sound as if they are professional versions.  If so, they will
likely also implement OOPS as does Borland's Pascal and will be fully
compatible with Borland's Pascal.  Hence, they will be almost equal to
C++ capabilities for a majority of apps.  Enough, from the sounds of it,
for my continued needs to upgrade my apps for OS/2 - in 32-bit Pascal.

You may wish to look into this if you have any on-going need for
upgrading any DOS or Windows based Pascal apps to 32-bits for OS/2.

ES>LR> Are you saying you converted all your existing DOS based apps to
ES>LR> C++ and OS/2 in one fell swoop?  Gulp.

ES>Gulp, indeed.  No, I still support old DOS stuff.  As new revisions are done
ES>however, they are written from the ground up for OS/2. Good programming
ES>practices in the old stuff means that this is not as daunting a task as it
ES>might seem, and with experience, it has become almost a routine exercise.

It is my understanding that last year a German magazine came out with
some modifications to Borland's BP 7.0 compiler that permit it to run
under OS/2 using PM - as 16-bit apps - a simple recompilation - no
conversion necessary.  I am not 100% sure, but it sounds like that might
satisfy your simulation memory needs, your need to multi-task, leaving
your apps with only 16-bit precison upgrade as might be necessary for
your apps.

A good step forward (IMHO) in comparison to converting your Pascal,
DOS based, apps entirely to C++.

I prefer to avoid conversions if at all possible - favoring
32-bit type recompiles for OS/2 . The lack of a 32-bit Pascal for OS/2
was bothering me.  I would rather convert Pascal procedural type apps to
Pascal OOPS, since that is significant for future development of my
apps in 32-bit Pascal for OS/2.

ES>One option you might consider is PL/I.  Although I know little about it, I
ES>understand that it is very Pascal-like, but more elegant.  I received a demo
ES>kit from IBM (which runs on OS/2), and though I haven't had much time to
ES>investigate it, it looks good so far.  I think this will be my next personal
ES>education project.
>>> Continued to next message


--- WM v3.10/91-0154