Subject: Performance of DecMate II Kermit (MS-DOS)
To: sy.fdc@CU20B.COLUMBIA.EDU
Various tests of MS-DOS KERMIT reveal some problems and limitations that should
be fixable in another release. Many limitations stem from the obvious departure
of MS-DOS KERMITs in general from the "sacred" PC-DOS KERMIT, but the DecMate II
demands a better fate considering how much is "built-in."
A) problems with hanging up. It appears that this Kermit can hang up with all
control characters failing to work when "get"-ing a file. Only <^C> does
anything, and that is to echo "^C<cr><lf>" and keep hanging 'til you do total
bailout hardware reset (with the internal SETUP interrupt key or more drastic
actions) and re-boot MS-DOS.
B) problems exist with REMOTE DIRECTORY command when used with IBM-PC-DOS KERMIT
Server at other end of wire local connection. The command does correctly give up
with <^C> bailout back to MS-KERMIT prompt, but it often makes server start to
create the temporary file $KERMIT$.TMP and then report " @-" as the entire
directory print out and give another prompt. If you quickly give another REMOTE
DIRECTORY command (while slow floppy-based PC-KERMIT server is still serving up
the temporary file), you can even get several null responses to more commands
(by the way, PC-KERMIT correctly reported the failure to send $KERMIT$.TMP).
These tests were conducted at 9600 and 1200 baud with no change; it also does
occasionally work. To confirm the hardware connection, the DecMate II was also
tested with K278 Kermit. That KERMIT had no problem "get-ing" multiple files
and was not flakey at all; switching to MS-DOS kermit produced consistently
flakey operation with this connection. CP/M KERMIT for DECMATE II also works
fine in "get-ing" files from the PC-DOS server. Since the path to all
communications is the 6120 (which runs K278 by itself), and the Z80 and 8086
use the same interface to talk to the 6120 when running CP/M or MS-DOS, it seems
the culprit is some problem with MS-DOS KERMIT software.
C) when erasing commands to the MS-KERMIT prompt, they don't backspace over the
eradicated characters (they act on it ok however, it's purely visual cosmetics
that are wrong).
D) server mode can also hang up when reversing the roles of server and served
KERMITs (also using Victor 9000 Kermit as the local KERMIT instead of PC-DOS as
local Kermit) with regard to <^C> "numbness."
E) MS-KERMIT for DECMATE II does respond to the setup mode for default baud rate
and to changes "on-line" induced by user SETUP changes in firmware/slushware of
same properly. This has limitations as a) it can't report the current baud rate
which can be figured out (if you're sneaky or have access to DEC's notes on the
subject) and you thus must use SETUP to find out what it is. b) the SETUP mode
only allows five popular baud rates, not the full complement of what can be done
from the 6120 point of view. I have hand-patched K278 Kermit for the 6120 alone
to arbitrarily do any rate the hardware is capable of, which is the usual collection of rates from 50-19200 baud. I don't know if MS-DOS can "tell" the 6120 to change the rates, but it would be nice if DEC told us how...
F) bells and whistles wish list. It would be nice to see some terminal enhancements to this Kermit, ala the PC-DOS version and its terminal tricks such as
buffering the screen to roll it back later, etc. This does not seem to
be anything so hardware specific that it can't be done by nothing more complicated than buffering the characters sent to MS-DOS's screen handler. The LK-201
DEC keyboard has more than enough doo-hickeys (read function keys, and
numeric pad extension keys, and the famous "gold" key, but I prefer to call them doo-hic keys).
Please note that in theory, this Kermit should terminally emulate just about anything the RAINBOW kermit can do, as they are similar environments with built-in robust terminal emulation provided by the firmware; it would be nice to see vt-220 emulation firmed up as in some other rainbow kermits.
All tests conducted on 256k XPU board DecMate II with rev. 11 roms and slushware
version 412 and hardware roms version 19N. The software is CU20B::KER:MSVDM2.BOO
de-"boo"-ed with MSBPCT.EXE on 17-JAN-87 on the DECMATE II; there is no MSBASIC
