To:   Distribution
         From: Keith Ball
         Re:   November 28 Constellation II meeting


         The first regular meeting for Constellation II was
         held at the Paragon building on Monday November 28,
         1983 from 1 to 3 pm.  The next meeting will be
         specified by me at later time.  It is based on the
         completion of action items by attendees.

         Phil Belanger handed out a preliminary version of the
         NIS specification.  His main action item is to complete
         it.  Main points consisted of a physical transporter
         driver, similar to the one on the Concept, and a socket
         manager to handle multiple protocols sharing a socket.
         A flexible interface usable by both single and multi-
         tasking systems to all functions of the two pieces is
         also specified.  The NIS is built to work with buffered
         or unbuffered and interrupt or non-interrupt driven
         transporters hiding hardware idiosyncracies from the
         rest of the system.  NIS is partitionable so a minimal
         amount of system resident code is necessary for small
         systems.  Performance concerns were raised for disk
         server protocols; the interface may need streamlining
         for normal system block I/O.

         Phil's other action item was to collect and record all
         device types used by the Constellation II lookup
         protocol.  We must assist him in specifying blocks of
         device numbers reserved for general device types.

         A standardization of the Constellation II boot process
         was discussed.  Denise volunteered to write a document
         describing the Constellation II standard boot procedure.

         My action items are developing a specification for the
         transporter driver for both buffered and unbuffered
         transporters, especially the IBM, and one for the disk
         driver changes to correct for the 8A transporter
         problems.  The transporter driver specification will
         also contain a buffer management algorithm.  It's
         definition will be based on the NIS specification for
         the physical transporter driver.

         We mentioned the importance of fixing the IBM transporter
         ROM to work with both the IBM and TI systems and fix the
         timeout bug in the current version.  However a hardware
         change is necessary to allow the software to be able to
         read the card's configuration.

         Bruce suggested the disk status command should be
         changed to reserve space for the host system to save the
         device's timeout period.  This could be used by the
         system manager to configure the host systems disk driver
         for different net environments.

         Command line driven user utilities (spool and mount
         manager) were considered necessary for the next
         Constellation II release.  If volumes on secondary disk
         servers are not mounted at boot time then this facility
         is a must.  Tri is currently working on the MSDOS and
         CP/M86 versions of the spooler.  It was suggested that
         Bill Kline could perform the mount manager changes.

         Denise suggested, for a later release, batch file driven
         capabilities added to the system utilities.

         We agreed we should develop and publish a Constellation
         II manual to go along with our Transporter Technical
         Reference manual.  The Constellation II manual should
         include a description of the two disk server protocols
         and the lookup protocol.  An OEM addendum should have a
         complete description of the all the Constellation II
         tables.

         Distribution

         Phil Belanger
         Sean Finley
         Bruce Kendall
         Denise Pitsch
         Steve Schwartz
         Mark Warshaw
         Bob Wohnoutka