[comp.windows.ms.programmer] Flaw in Windows design?

mmshah@athena.mit.edu (Milan M Shah) (11/01/90)

I would like to bring up a new topic of discussion; I think there might be
a serious flaw in Windows as we know it.

I refer to the lack of installable device drivers *for windows*. I know
that we can communicate to devices by using DOS to load the device drivers,
and then try and use the device driver interface from Windows apps. However,
there are at least two drawbacks to this:

1) Windows itself knows nothing about the device driver. A glaring example
   is the hard disk - the code to access the hard drive is hard coded into
   windows (I think), and the only thing you can do is turn virtualHDIRQ off,
   in which case it goes through BIOS. Would it not have been better if
   instead there were a well defined device driver interface, and windows
   came with a default one? That way, people could get a device driver for
   their whiz bang HD and get super performance out of it.
2) DOS device drivers take up space in the lower 1 Meg. This takes off
   space from every dos window I run. Consider the case of a windows device
   driver for a network card - if I loaded this "device driver" while
   initializing windows, I wouldn't need to have the device driver take up
   space in every DOS window. Additionally, windows apps would be able to 
   properly share the resources etc.

In general, I look upon windows as an operating system (what else do you
call something that provides memory, time and other resource management?).
But I have never heard anything about an *installable* device driver for
windows - true, there are the standard drivers like mouse.drv and sound.drv,
but nothing about installable device drivers for any arbitrary piece of
hardware.

But then again, perhaps I am missing something big - perhaps the DDK does
allow you to write device drivers for your own piece of hardware.

Comments?

Milan.