"CP/M compatible" vs. "MS-DOS Compatible" machines?

[Dave McGuire]

On Jan 29, 2008, at 4:33 PM, Joshua Alexander Dersch wrote:
>> In the early-mid 80's a program was "well behaved" if it did it's  
>> I/O thru DOS calls. Those programs would run on just about anything.
>
> Were there similar problems in the CP/M world?  That is, was it  
> commonplace for there to be CP/M programs that bypassed CP/M BDOS  
> calls and wrote directly to a specific machine's hardware?  Seems  
> like CP/M developers were more disciplined in this fashion, but  
> maybe it's just because in the CP/M arena there were so many  
> different pieces of hardware it was the only way to do it?   
> (Whereas with IBM, the PC was seen as more of a reference standard,  
> even if it wasn't really that way in the beginning?)
> I'd be interested to hear opinions from people who were there at  
> the time, since it was a little before my time.

   I was there; I was a CP/M weenie for many years.  I still use it  
from time to time; it's lots of fun.

   The CP/M BIOS definition doesn't provide for all the things a user  
might want his/her hardware to do.  Accessing serial ports  
efficiently is one thing...Some popular CP/M communications programs,  
such as the MODEM7 family, used the concept of "overlays" (not the  
kind that we use in the PDP-11 world!), which are assembly language  
modules which are written for your specific hardware that present a  
unified (BIOS-like!) interface to the rest of the MODEM7 code.  It  
was intended that end users would write these routines for their  
specific hardware.  (in a day when end users were typically highly  
technical people)

   Formatting floppy disks is another example.  The CP/M BIOS has no  
provision for floppy disk formatting, so a formatter program had to  
be written for each floppy controller, and it accessed the  
controller's registers directly to perform the formatting (and  
sometimes verification) function.

   Other than specialized hardware (lab I/O, speech synthesizers,  
real-time clocks) there weren't many instances of programs directly  
accessing I/O in my experience.

            -Dave

-- 
Dave McGuire
Port Charlotte, FL