>Andreas Freiherr wrote:
> Hello Stuart,
> welcome to the list from me, too. I am also fairly new to the list
> (joined only a couple of months ago), and have some similar interests.
> Perhaps I should take the chance and put together a few details about me
> as well.
Jerome Fine replies:
Hello Stuart and welcome to the list from the software side. I
think there are very few whose interest is almost totally software,
but a sprinkle probably does exists.
> The 6502 was the first microprocessor that I learned details about, but
> the 6809 gave me the actual kick. However, I have never owned one of the
> home computers that use any of these (nor the Z80). Instead, I have a
> boxfull of chips that were soldered together by two friends, containing
> a 6809, and I started making a rackmount 6809 several years ago, but did
> not finish this project yet (if you can ever call such a project
> "finished" at all).
I can't say that I every learned very much about hardware. The
software was always my cup.
My first system was an IBM 650 which only has assembler
language and NO operating system. One user at a time
on a system which occupied a whole room.
> The PDP-11 infection happened when I did my diploma thesis: a
> temperature measurement / acquisition system on RSX-11M. Today, I have a
> number of Q-bus and UNIBUS systems in working order at home.
I have a number of PDP-11 Qbus systems as well, but the Unibus
systems are BOTH too heavy and used to be too expensive.
Plus, I focus on RT-11 operating systems, fixing bugs and
making enhancements. My current goal is to produce a
Y2K/Y10K V5.03 of RT-11 for hobby users. NO ONE
else seems interested, so I will just putter along.
> For RT-11, I earned a reputation as an "overlay expert" when I managed
> to run several applications based on our own GKS (Graphical Kernel
> System) library within the address space of 64kB. The overlay tree was
> ported from RSX, and the same library was available for TOPS-10, VMS,
> several PC (DOS) FORTRAN compilers, and one FORTRAN implementation on
> the Atari ST.
NOW that would be very interesting. Is there a copy of the source
available? Was it ever submitted to DECUS?
About 10 years ago, I ran out of address space on one of the applications
in an RT-11 environment (obvious on a PDP-11). In order to extend
the available memory, we decided to allow one of the cardinal rules to
be violated when overlays were being used. Since this was under RT11XM
and already 64 KBytes were being used of address space along with
virtual arrays of over 2 MBytes, it seemed like the best solution. Yes,
for those of you who know RT-11 under RT11XM well - when VIRTUAL
arrays are being used, 8 KBytes of the I/O page are lost to the use of the
Memory Management Hardware.
In any case, the cardinal rule that was violated was the an overlay in
the same address space could NOT call an overlay in the same address
space or memory region. There is an excellent reason of course - the
return address code is not available when that happens - sort of obvious
if one thinks about it. Plus it would tend to create thrashing behavior
in a low memory overlay situation when the operating system would
continually need to read the different overlays back into memory
each time. However, when virtual overlays are used, the code always
stays in memory and the operating system just maps in (and out) the
various portions of the code. One additional rule that also must be
imposed when the above cardinal rule is violated is that any arguments
passed during a CALL to another overlay should probably be in the
root so that the new overlay can "see" them.
Fortunately, references to VIRTUAL arrays are always done via the root.
In any case, to make a long story short, the overlay handler was easily
modified to save the overlay number on the stack and when a different
overlay region was used for the new overlay, the overlay handler then
automatically restored the old overlay region without the user having to
do the extra code that others proposed to handle that situation.
Multi-user Basic (MUBASC) does this all the time, but users are not
aware.
> I have been using VAX/VMS since some V4.n version, around mid-80s, and
> OpenVMS/AXP since 1992, when I had the pleasure to run a field test site
> for V1.5, clustered with VAX/VMS V5.5. One of my projects was writing a
> print symbiont (using the SMB$ interface to the queue manager, not PSM$
> routines).
I used a VAX/VMS system in the early 1990s and discovered all
of the wonderful enhancements. Ever since I have attempted to
figure out how to port some of them back down to RT-11. One
such as the Logical Name List capability is close to completion
as a Path Handler in RT-11. Also similar to the PATH NAME
is DOS.
In any case, welcome to the classiccmp list. There are few other
software addicts such as myself.
Sincerely yours,
Jerome Fine
--
If you attempted to send a reply and the original e-mail
address has been discontinued due a high volume of junk
e-mail, then the semi-permanent e-mail address can be
obtained by replacing the four characters preceding the
'at' with the four digits of the current year.
Received on Sun Jan 26 2003 - 11:34:01 GMT