The reason programmers wrote DOS programs this way was that under DOS every
program "owns" the entire machine (so doing direct hardware manipulation
was a totally proper thing to do), and directly manipulating the hardware
is often the only way on older, slower PCs to get acceptable performance.
(Now we have much more capable PC hardware, so we can get the same sorts of
sluggish behavior by having huge programs that run every action through an
equally bloated operating system--but that is another issue. <grin>)
The issue of drivers for NT is being addressed by Microsoft by their
integration of the two kinds of Windows (9x and NT) for driver-writing
purposes. This applies, however, only to Windows NT 5.0 and later. At that
point we "should" be able to use the same drivers for all our devices under
both Windows 98 and NT 5.
You may think that everyone who is running Windows 9x has long since
upgraded away from all those bad-old DOS programs. That is absolutely not
the case! If you have a program that does some tasks you must do and it has
never been upgraded, but you also have to do other things that require
modern Windows programs, the usual approach is to use a Windows 9x machine
(or for many people, a DOS + Windows 3.x machine) to accomplish both tasks.
Moving this sort of user to Windows NT is going to be at the very least a
challenge.
For now I think the only way to do it is to be sure you use only the old
FAT16 file system on the C: drive, and set up a dual-boot configuration so
you can switch to real DOS when necessary, and then reboot back into
Windows NT at other times.
John
At 07:14 PM 7/2/98 -0700, you wrote:
>John, what is the basis of you experience with DOS programs refusing to run
>under NT? In the 4 years of extensive use of NT I have only a very few apps
>that do not actually run under NT, and those that require direct hardware
>interaction. Make no mistake NT _is_ more stable. However the price to pay
>is that you cannot interface directly with the hardware you have to go
>through the API/HAL to get there and that can limit you. That is why
>precious few Scanners and similar hardware work under NT. It is improving
>but only slowly. Part of the Problem is the API set and the increased
>complexity in working with NT, but most of the device support problems are
>hardware vendors just not supporting NT, that is changing but way too slowly
>and it will only speed up if M$ drive the process far more aggressively that
>they have to date.
>
>Phil
=======
John M. Goodman, Ph.D., author of "Peter Norton's Inside the PC,"
Seventh Edition (Sams 1997, ISBN 0-672-31041-4)
=======
![]() |