Well, it took a while, but I now have the HP9836 Kermit working well
enough so I'd trust it at other sites.  This version is intended to
work only on the HP9836, so it is heavily dependent on the HP pascal
extensions, mostly the module facility.  It probably will not be easy
to port to non-HP machines.  The actual Kermit protocol routines are
from the RT-11 Pascal version from Ontario/CU.  The user interface has
been completely rewritten - one of our requirements was that it be
reasonably idiot-proof.

The current version is minimal, and at this point is really a
prototype.  It will only transfer text files, only one file at a time,
and the error handling is not the greatest, but it works for
everything it's supposed to do.  I plan to be adding a lot to this
implementation quite soon, to improve the user command interface,
improve error handling, add login packets, and maybe binary transfers.

The important features (marked by +) and misfeatures (marked by -) of
the current version are

	- errors not handled gracefully
	- only transfers one file at a time
	- does not handle wild cards
	- only transfers text files
	+ continuous status display during file transfers
	+ can talk to a server
	- does not handle timeouts
	- only acts as local Kermit
	+ reasonably friendly command interface, modeled after TOPS-20
	  COMND facility.

I have made the sources and documentation available on RUTGERS.ARPA in
<Gallaher.Kermit>KRM*.TEXT. There are six source files, which contain
about a dozen modules altogether.  They use only the recommended
library routines and such that HP distributes.  The file KRMDOC.TEXT
summarizes the (mis)features and gives details on how to run this
Kermit.


Mike Gallaher
Gallaher@Rutgers.Arpa
23-Jan-84 11:34:09-EST,2199;000000000000
Date: Mon 23 Jan 84 11:34:09-EST
From: Frank da Cruz <cc.fdc@COLUMBIA-20.ARPA>
Subject: New release of KERMIT for HP98xx
To: Info-Kermit@COLUMBIA-20.ARPA

The files are in KER:HP9*.*, accessible via anonymous FTP from host
COLUMBIA-20.  The problem he alludes to is due to an omission in the
KERMIT Protocol Manual, which will be fixed in a new revision: data
fields of all packets EXCEPT the Send-Init (S) or the Info (I) packet,
or the ACK to those packets, or the File Attributes (A) packet, should
go through the encoding and decoding mechanisms for control prefixing,
etc.  - Frank

----------

Date: 20 Jan 84 05:29:39 EST
From: GALLAHER@RUTGERS.ARPA
Subject: New HP Kermit
To: cc.fdc@COLUMBIA-20.ARPA

Frank,

I have a new version of HP-Kermit.  All the source files have been modified,
and there is one new file, KRMVERS.TEXT.

I was going to add a few more things to this version of HP-Kermit
before sending it over, but I discovered a severe bug in the last
version that had to be fixed.  It seems that the old protocol code did
its control unquoting in the receive packet routine.  Not only did
file data packets get unquoted, but ALL packets did!  This did not
cause a problem until 20 Kermit started setting the QCTL and QBIN
fields in its send-init packets to #Y.  Since # happened to be the
same character HP kermit was using for local quoting, HP kermit took
the #Y to mean "use ^Y as the control quote"!  I fixed it so the
unquoting is only done for data packets.

Since I lifted the protocol code from RTKERMIT, this bug is almost
certainly in that version also.  If you want I can try fixing it in
there too, but I have no way to test it.

By the way, I couldn't find any mention of this potential problem in
the protocol document, though I didn't look very hard.  I admit that
quoting is obviously meaningless when the quote characters have not
yet been defined, but some implementations (like this one) might not
account for that...

Besides the bug fix, the new version has nicer error reporting, the
code in the protocol module is a little cleaner, and responds to
interrupt keys ^X and ^Z.

Mike (Gallaher@Rutgers)
-------