jgray@toad.pilchuck.Data-IO.COM (Jerry Late Nite Gray) (11/10/87)
<>
I was just reading an article in the November 1987 issue of IEEE Software
titled, "Living in the Next Generation Operating System". It's not about
OS/2. In fact, I have been wondering about which generation (acording to
this articles definitions) OS/2 is a part. This is an open question to
anyone who knows.
In this article it describes three generations of operating systems (OS)
(First, Current and Next). Though there are many distinguishing features
of various OSs the distinguishing criterial for the three are:
First: The intermodule(task/job)-communication interface was normally
specified in a special purpose job-control language that mapped
logical input and output names used in a module to physical files
and a sequence of modules to be run. Each module was processed to
completion before the next was initiated, and the processing state
was externalized in the contents of the closed file transmitted
between phases.
Under this rather basic limited description we can file both DOS and
VMS though they are very different.
Current: This is a logical extension of the sequential phases paradigm to
interactive applications:pipelining. Each phase can begin processing
information before the previous phase completes. Intermediate data
(between phases) is not buffered in files but is consumed as it is
produced.
This generation had its roots in two insights. The first is coroutines,
which eliminated the tyranny of subroutines' master/slave relationship
and allowed the retention of local state (both control and data) as
control was passed back and forth. The second insight was that the
normal I/O interface to the operating system was itself a candidate
for such coroutining. The operating system could act as a wiring
intermediary between modules that let them be pipelined through a
conventional, but now virtual, I/O interface.
Unix, of course, defines this generation of operating system.
Next: In many ways this generation is a natural extension of the previous
one. specifically:
* Transient data can become presistent data. Rather than pushing the
data around through pipes where it is processed as it flows past,
it is held in one or more repositories where it can be accessed
(shared) by many modules.
* Point-to-point communication became broadcast communication. This
let any module interested in a change respond to it, thus extending
the implicit dataflow control structures of the pipelining paradigm.
* Intermodule interfaces became data-base interfaces.
* Character strings became structured objects. Using a database as a
repository for the data being communicated between modules lets the
full power of its schema be used to structure that data.
The articles example was an OS only referred to a the "prototype"
which they had been working on for over three years. Their "charter"
was to research and develope mechanisms for automated support of
software development which they have been persuing for over 15 years.
The OS was a result of needing an extended database operating system
with extensions for abstract data types, object-oriented message
passing and multiprocessors.
I actually know very little about OS/2 and would be interested in how one
could clasify it given the above definitions.
-------
Another interesting thought about OS/2 (unrelated to the above). Has it occured
to anyone that DOS is the most widely used piece of software by the PC computer
illiterate? I know people to "boot up" some application in the morning and
stay in it till they shut the power off an go home. The PC is just some
non-black box that performs an office function. With the growth of PC usage
a vast majority have never see early releases of DOS. Some of these people
are going to get caught up in the mad rush for OS/2........and pay for it.
---------------
Jerrold L. Gray
UUCP:{ihnp4|caip|tektronix|ucbvax}!uw-beaver!tikal!pilchuck!jgray
USNAIL: 10525 Willows Road N.E. /C-46
Redmond, Wa. 98052
(206) 881 - 6444 x470
Telex: 15-2167sam@ncsuvx.UUCP (11/20/87)
Because of a need here at the computing center I wrote a VT100 emulator. The need was for a VT100 emulator that truely supported the wide and wide/high character sets of the VT100. I wrote the emulator to run in CGA HIRES mode 6 because that is the most widely used mode/card on campus. I have font editors that are easy enough to create the fonts it uses. I would like to make the program allow changing of the fonts at run time. Currently the font editors dump include files that assemble in. Okay so get to it? I have a lot of things that I want to add to it. I would like to add Hercules and EGA support. I would like to add ANSI graphics capabilities or emulate another terminal like the TEK's. My problem is whether there is a need for another VT100 emulator or one that runs in graphics modes and supports extended graphics features. File transfer is also something that would be nice to add. If there is anyone that is interested in it or is interested in future needs in a PD emulator or wants to help drop me a line. Also, I have seen references to a Kermit that emulates a TEK terminal. Is this true? There has also been talk of a version of Kermit greater than 2.29b. If there is one out there is it the new C version? Thanks, I think. Sam ======================================================================== = Sam Moore = = Computing Center - Systems UNC-CH Should Hire Earl Bruce = = North Carolina State University = = sam@ncsuvx (internet) ====================================== = sam@ncsuvax (bitnet) = = samm@ncsuvm (bitnet) = ===================================