rick@cs.arizona.edu (Rick Schlichting) (01/14/91)
--------------------------------------------------------------------------
Trip Report -- Kyushu University (May 28, 1990)
Richard D. Schlichting (rick@cs.arizona.edu)
Associate Professor
Department of Computer Science
The University of Arizona
Tucson, AZ 86721, USA
[This report represents the personal opinion of the author, who was on
sabbatical in Japan from Dec. 1989 through July 1990. The sabbatical
was supported in part by grant INT-8910818 from the NSF U.S.-Japan
Cooperative Science Program.]
---------------------------------------------------------------------------
On May 28, I visited Kyushu University, which is located in
the southern Japanese city of Fukuoka. I was picked up by my host,
Prof. Kazuo Ushijima, at about 9:30 AM. We then walked the 10 minutes
required to reach his office, entering the campus through what appeared
to be a small south gate. Upon arriving, we chatted for bit
and then he briefly showed me a PC-based Japanese writer's
assistant. The first version was written in a version of Snobol4
that they had augmented with Japanese features back in 1983 or
so, although the newer version he demoed is written in C. The former
version was of special interest to me since one of the developers
of Snobol4 -- Ralph Griswold -- is a faculty colleague at the University
of Arizona.
After a few minutes, Ushijima passed me on to the Associate
Professor in his laboratory, Prof. Keijiro Araki. After a few of the
usual pleasantries, I asked about equipment, and he ended up giving
me a very complete rundown and tour of the facilities. They are, in
short, impressive. The bulk of their equipment is from DEC, and they
are apparently one of DEC-Japan's most important academic customers
(the president of DEC-Japan was going to visit a couple of days after I
left.) The centerpiece of their equipment is a Vax 8800, which is used
for both departmental research and teaching. This is, in fact,
the first of this type/class machine I have seen on my visits;
they acknowleged that this is unusual, although claimed that it
was common for departments to have 780-class machines several years
ago. They also have 30 Vaxstation 2000's that are used primarily
by undergraduates, mostly to get windows on the 8800 (seems that their
lack of memory--only 4 MB--makes them very difficult to use
as real workstations.) Another 11 Vaxstation 2000s were distributed
around to different laboratories for use by graduate students.
There are also 3 or so Vaxstation 3500s, as well as some Xerox
machines for word processing. In Ushijima's laboratory specifically,
I also saw 7 or 8 Suns, as well as a scattering of Macs, PC-class
machines, etc.
In addition to a tour of the facilities, Araki also gave
me an overview of the personnel in the lab, some information
on the department, and also a feel for how the department fit
into the University. Among the interesting points:
-- The department is called the Dept. of Computer Science and
Communication Engineering, having evolved from the Communication
Engineering department.
-- The department historically has had 7 chairs, with 2 being
devoted to CS (Prof. Ushijima's chair is called Computer
Software.) However, they now have 11 chairs; two of these
chairs were from other places in the University, while they
received one each from the Electricity Dept. and the Electronics
Dept. (two dapartments with which they closely interact.) The
new chairs are partly a reorganization--bringing together
AI-related work--but also some growth.
-- In the lab currently are 3 PhD students, 7 MS students,
and 10 undergrad. This is apparently a bit lower than
normal as he mentioned that they had more students last
year. He also said that this was about average for the
other labs in the department.
We were also able before lunch to talk some about his own
work, which focuses primarily on formal semantics of concurrent
and distributed programs. Despite his use of formal approaches,
his goal--like mine--is to develop pragmatic techniques. One
thing that he has done is to attempt to apply some different
techniques to a robotics system. The work is being done jointly
with with people from Yasukawa Electric works, which is located
in Kita-Kyushu; this company originated as a manufacturer of
heavy machinery for steelbuilding, chemical plants, etc, but
has recently moved into the manufacture of industrial robots.
The example program itself is a robotics application in
which they wish to establish properties about the movement
of the robot. For example, they want to argue that synchronization
is done correctly so that the arms never touch each other. The
first attempt was done using temporal logic; the basic logic was
from Kroger's book, although they found it necessary to do
extensions. The second was using the algebraic approach developed
by Manfred Broy, a natural since Araki had recently returned from
spending 10 months on a Ministry of Education grant at Passau, West
Germany studying with Broy. He gave me several papers describing
this work, mostly in Japanese.
At this point, Araki introduced me to Dr. Norihiko Yoshida, a
research associate working in distributed systems, and Zengo Furukawa, a
lecturer associated with the group, but technically from another
department. We talked about research, with Yoshida and Furukawa
giving me an overview of their work. Given my background, Yoshida's work
was more interesting to me. Basically, what he is trying to do is
implement distributed virtual shared memory for an AI-type blackboard
application. The specific novelty of the work according to him is that
the implementation is based on point-to-point "request" messages rather
than a broadcast algorithm; the rationale is that this is supposed to
work also in computing environments that are not based on a LAN.
That claim seems reasonable, although it was not really clear to me
exactly how his work relates to research in the area of reliable
broadcasting or other work in DVSM. He gave me a paper in English
describing the work.
Following lunch and my research talk, we adjourned to Araki's office
for a discussion among Araki, myself, and Dr. Jingde Cheng, the other
research associate in the lab. He was been working on several things,
including an event monitor and debugger for Ada programs, formal methods
for detecting deadlocks in Ada programs, and on non-monotonic logics.
The work on the event monitoring was especially interesting to me
since its use of partial ordering of events makes it very much related our
own work on the Psync IPC mechanism. We also chatted for awhile
about problems with Ada before I had to leave to catch the Shinkansen
to Hiroshima.
Address
--------
Department of Computer Science and Communication Engineering
Kyushu University
Hakozaki, Higashi-ku
Fukuoka 812, Japan