[time-nuts] Lady Heather's Leap Log

Lux, James P james.p.lux at jpl.nasa.gov
Thu Jan 1 14:02:17 EST 2009

On 1/1/09 8:37 AM, "Chuck Harris" <cfharris at erols.com> wrote:

> Lux, James P wrote:

> The fault was in the thinking that there would never be more than one or
> two terminal types used as consoles and I/O devices, so the applications
> programs should handle I/O directly.
> That fault was fixed once and for all with unix's termcap... the idea that
> the driver should present as common as possible of an interface to the
> application.

MSDOS was a microcomputer OS and inherited most of the features of other
microcomputer OSes at the time.  Termcap was a royal pain in the rear to
deal with, and consumed significant (scarce) resources on machines where 32K
of RAM was the whole load. I can't fault MS for the choice they made back
then. And, in a small resource system that's talking to a glass TTY as a
console, why not just code for what's out there. (after all, all you had
between your application and the hardware was a BIOS call.. CP/M provided
almost nothing between file and console, e.g. PIP)

Granted, by the time they were building "real" operating systems (say, from
NT onwards) they should and did have a more sophisticated scheme. But by
that time, lots of folks had invested in compatibility with the earlier
MSDOS,CP/M style stuff.  And, besides, MS was already heading towards
windowing environments where the OS would manage the video display directly
with bitblts etc, rather than worrying about supporting hundreds of
different serial console hardware and terminals.

I remember looking at various Unix implementations in the mid 80s for
microcomputers (remember Charles River Data Systems, for example), as well
as Unix-like implementations (Cromix).. My roommate at the time was toiling
on 68k systems, including porting the original SUN workstation stuff to
another 68k platform, and I well remember his comments that the curses
package (which uses termcap) was well named. I think there were also some
licensing/copyright issues with the plethora of Unixes around at the time.

> MS had ample opportunity to follow that example, as it was well established
> back in the early 1970's.  They chose instead to reinvent the wheel, and
> ignore
> most of the advances that came before them.

Not really.. I'd venture that MS has NEVER aimed their OS towards supporting
serial consoles in any serious manner, so why would they care about termcap
and all it's associated cruft. They've always worked in terms of a memory
mapped display of some sort (viz the horrible serial port support in the IBM
PC BIOS and Int14 in MS-DOS).

> They still do.

As do lots of other folks. You choose what works for you in your situation
as you have it. Sometimes, a Unix derivative works best, other times,
something else does.

> -Chuck Harris
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.

More information about the time-nuts mailing list