COMMENT ⊗   VALID 00006 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002
C00005 00003
C00008 00004
C00011 00005
C00013 00006
C00015 ENDMK
C⊗;





This note deals with the  relationships between the LCDS, the  Alpha-micro
and the Futuredata development  system. The original  note was written  on
December  29,  1978.  Since  that  time,  I  have  discovered   additional
information about the Futuredata system; that information, which conflicts
with earlier claims, is analyzed in an addendum.



			       INTRODUCTION



It should be made  clear that the Alpha-micro  machine is not a  prototype
LCDS, either in terms  of its hardware  or its software.  Rather, it is  a
machine which  possesses  sufficient power  and  flexibility that  we  can
develop software ideas which  can be incorporated  into the LCDS  machine.
These ideas include:

1. high-bandwidth display interaction; for example, memory-mapped displays
with multiple  windows for  editing,  debugging, and  monitoring  multiple
tasks.

2. the development of a Pascal dialect, suitable for implementing the LCDS
operating system and user programs.

3. the merging of the ideas expressed  in 1) and 2) to give an  integrated
interactive development system for a high level language.

Please note that all  of these activities can  proceed in parallel on  the
single Alpha-micro.  Even if we assume  that a Futuredata station had  the
appropriate tools, we would have to supply at least three such stations to
maintain the same flexibility: one for  the display projects, one for  the
language project, and one as a Network Control Processor.

From the  material available  to me  (on 12/29/78)  I concluded  that  the
Futuredata system  neither  supplies  3) nor  gives  sufficient  tools  to
support 1) or  2). The next  sections discuss the  rationale behind  these
conclusions.








			 OBJECTIONS TO FUTUREDATA(12/29/78)


First, the Futuredata display only supports a traditional 80x24  character
display. That is an insufficient size to support complex interaction.   We
must be able to investigate the use of larger "windows" --of the order  of
4000 characters-- with the intention of including such display  philosophy
in the LCDS  and its successors.  Such displays have  been used for  years
within the research community of Stanford, MIT, and Xerox PARC. The  added
flexibility of such  display systems  has long been  appreciated by  these
university groups;  however, the  flexibility need  not imply  complicated
operation.  The Xerox  Learning Research Group  uses their systems  rather
dramatically in educating  8-12 year old  children. The key  is the  human
interface; to be successful we need to develop an interface philosophy for
our development system. The Futuredata does not supply the components.

An integral part of the display is the rapid presentation on data. I would
expect that major revisions of the Futuredata operating system would  have
to be made for such performance  to be obtained.  It is questionable  that
we would be granted sufficient to  the internal workings of the system  so
that such surgery could be performed. The Alpha-micro is a general purpose
system; such operations are allowed.

Finally, the  Futuredata   literature  says   that  "editors,   debuggers,
assemblers, and compilers" are available;  however no details about  which
compilers "exist" are given. Pascal is a must; UCSD Pascal is supplied  by
Alpha-micro; is it supplied by  Futuredata? For further questions see  the
next section.









				QUESTIONS(12/29/78)


The literature on Futuredata is quite sketchy. Below are several questions
which came to mind.

1. What languages are supported? For  example, is there a Pascal  dialect,
and if so, which one?

2. The floppy  system guarantees file  response in less  that 10  seconds.
That is totally unacceptable. Do they offer a hard disc option?

3. One article quotes $4975 for  a 6802-based system; another says a  "per
station cost of $7694". Which is right.  What is included in the $7694?  a
floppy?

4. In the ELECTRONICS  article, they mitigate some  of the drawbacks of  a
"star system" by saying that any AMDS can replace the NCP. However, it the
NCP fails then something must be  done to reconfigure the system. What  is
that process?  For  example,  does  every  AMDS  always  contain  the  NCP
software, or  must it  be loaded  from an  external source?  (ah,ha!!  see
addendum of 1/3/79!)

5. The AMDS  comes with  several different host  processor options  (6800,
8080, 8086, for example). Can you use  an AMDS whose processor is of  type
"X" to debug  a processor of  type "Y"?  For  example, let X  be 6802  and
Y=8086; that is,  do they  supply some emulation  software? Otherwise  the
"Universality" of their systems is of the trivial kind.









			       CONCLUSIONS


There are really only two conclusions:

1. The Alpha-micro is not a prototype  of the LCDS, either in function  or
hardware. The Alpha-micro is a development tool for producing new software
ideas.  As such, we need generality and flexibility. We do not want to  be
constrained by the design decisions of another manufacturers product.

2. We  can  do  better  than the  Futuredata;  much  better.   Unless  the
Futuredata articles  are being  very modest,  there is  little novelty  in
their design.  We can prototype a  superior system on the Alpha-micro  and
transfer the ideas to the LCDS.









			  NEW YEAR'S REVELATIONS


After writing the previous note, I  received a couple of December's  trade
magazines. Both had Futuredata information. December's MINI-MICRO  SYSTEMS
shows the  single  station cost  up  to  $10,500 (from  $7694  in  August)
However, the December  15 EDN was  even more illuminating.  They list  the
following costs:

		     $4850 satelite terminal
		     $5350 network control processor
		     $4950 disc drive
		     $2650 line printer

The most interesting price is that of the NCP. The control processor is no
longer just any AMDS, but is a special unit. If it fails, the whole system
goes. It also says that a single station system costs $17,800.