June 30, 1991

By Mike Bilow, N1BEE,
For GRINOS/GRINOS-S KA9Q 910618/PA0GRI 910625v1.7a/N1BEE 910630v0.4á


     All users of GRINOS/GRINOS-S are invited to join the grinos@n1bee
mailing list.  This is what I use to distribute advisories and updates.
There is no requirement that you actually use GRINOS in order to join the
mailing list.  Just send a message to n1bee@n1bee and you will be added.


     I. KA9Q/PA0GRI changes

     This release of GRINOS uses KA9Q 910618 as its base instead of 910605.
Most of the changes are of little or no importance to ordinary users.  The
experimental Kantor model for Internet wormholes between NET/ROM points is
now implemented using Anders Klemets' code as the "axip" interface.  Since
no one is going to use it locally, I would have taken it out, but there is
no way to do that easily, so we are going to have to let it consume memory
for the time being.

     There are extensive bug-fix revisions to the mailbox code.  The local
chat functions should now work when initiated at the mailbox.  (This was a
result of a timer being set for two minutes instead of two seconds.)  Users
of the forwarding functions that allow mail handling between NOS and W0RLI
or other conventional packet BBSs will also see major changes.

     KA9Q seems to have taken another stab at getting TIP and PPP to work,
although I would consider them highly suspect.


     II.  N1BEE changes

          A.  Bug fixes and default modifications

     The ttylink screens have been fixed.  Two users of ttylink in this
version of NOS who are communicating will see identical top windows on
their screens, except that each will see outgoing text in high intensity
and incoming text in normal intensity, which was the original intent.
The problem may actually have been associated with a documented bug in
the Borland C++ 2.0 compiler run-time libraries.  It has now been fixed
by explicitly calling clreol() at the necessary points.  (Reported by
KA1MF and WA1OAJ.)


          B.  New features

     I have extensively modified the "ax25 heard" functions.  Because of the
very large amount of memory consumed by the logging of calls heard, it is
important to minimize waste.  Wide area stations, especially switches, pick
up error frames quite often that have corrupted callsign fields.  Issuing
the "ax25 heard" command (or the J command from telnet) on the wb6nil switch
will dump pages and pages of such junk calls, all of it using up memory.
I have now implemented code that prevents calls from being added to the
heard lists unless they consist entirely of upper case letters, digits, or
spaces.  This simple filtering eliminates about 90% of the junk.

     Unknown to most users, NOS maintains a log of callsigns to which frames
are directed, in addition to those from which frames are transmitted.  One
good reason that this is not generally known is because there is no way to
access this list, and it is never actually used for anything.  I have enabled
a new command "ax25 hearddest" that dumps this list on the display in a format
similar to that used by the "ax25 heard" command.

     Still, most users probably do not need the heard lists at all.  Because
there are future plans and stubs of routines to use the lists for routing
information, I have decided not to disable this feature.  I have added a
new command "ax25 filter" that takes a bit mask argument that will inhibit
the logging of calls heard, but this command defaults to logging of both
source and destination calls enabled ("ax25 filter 0") for compatibility.
Users can disable either logging of source calls ("ax25 filter 1"),
destination calls ("ax25 filter 2"), or both ("ax25 filter 3").

     As an accidental side effect of this new command, the shorthand "ax f"
command will now map as "ax25 filter" instead of "ax25 flush"; users of the
"ax f" command should type at least one more letter to correctly access the
old "ax25 flush" command.


     III.  Future plans

     KA1MF suggests allowing commands which require the address of a control
block (such as "tcp kick" or "tcp status <tcbaddr>") to take a four character
short form of the address, since the memory allocator guarantees that the
last four charcaters are always "0008" anyway.  This sounds like a good idea,
and it will be in the code as soon as I can get to it.