Meeting Agenda Meeting Agenda The term ``metamedia'' refers to a second order media form that can be used to specify, describe, and implement other communications media. Computer software is the prime (and most flexible) example, but as usually implemented is rather inconvenient in this context. The introduction of Java goes a long way towards turning computer software into a usable medium, but the HotJava approach results in isolated monolithic applets. Rather than providing ``applets'', Java objects should be used to describe language semantics, i.e. to create specialized media on the fly.
The MetaMedia project is looking at combining SGML and Java to provide WWW documents with arbitrary structure and arbitrary semantics. Our architecture aims to extend the object-oriented structure of Java to the document definition level, to provide a flexible system for experimenting with different ways to structure "executable content".
In this talk, we first discuss this untapped potential and then show how it can be used to better advantage for a significant category of large real-time applications. Specifically, we describe a new object-oriented modeling method called ROOM (Real-Time Object-Oriented Modeling). ROOM partitions the domain of software phenomena into several (formally correlated) levels that are differentiated on the basis of scope. Different specifi- cation languages are used for each scope. For specifying implementation-level detail, ROOM relies on conventional object- oriented programming languages, such as Smalltalk or C++, but uses a formal graphical notation for specifying the higher-level aspects of software. These higher level capabilities include the explicit and formal definition of horizontal and vertical (lay- ered) object structures as well as the description of concurrent behaviour using a variant of Harel's statechart notation.
Biographical Sketch:
Bran Selic is the Vice President for Research and Technology and co-founder of ObjecTime Limited, a Kanata-based company special- izing in development tools for distributed real-time software systems. He has over 20 years of design and project management experience in real-time software development covering a wide range of applications including telecommunications, aerospace, and robotics. He is co-author (along with G. Gullekson and P. Ward) of the book "Real-Time Object-Oriented Modeling" (published by John Wiley & Sons, 1994) which describes the ROOM methodology for real-time system development.
Date: Wednesday, September 27, 1995
Time: 6:30 PM
Place: MC 3006
DEPARTMENT OF COMPUTER SCIENCE
UNIVERSITY OF WATERLOO
SEMINAR ACTIVITIES
MASTER'S THESIS PRESENTATION
-Thursday, September 28, 1995
Keith Pomakis, graduate student, Dept. Comp. Sci.,
Univ. Waterloo, will speak on "Reachability Analysis of
Feature Interactions in Service-Oriented Software
Systems".
TIME: 1:30-2:30 p.m.
ROOM: DC 3301
ABSTRACT
A feature in the context of a service-oriented software
system is a program module that is added to a basic
service in order to add functionality. An example of
this type of system is the telephone system, where
customers can subscribe to add-on telephony features
such as Call Waiting, Three-Way Calling, and Call
Number Display. An inherent problem of such a system
is the feature interaction problem. A new feature
interacts with an existing feature if the behavior of
the existing feature is changed by the presence of the
new feature.
This thesis describes an approach to detecting feature
interactions during the requirements phase of feature
development. The approach involves specifying features
in the context of a layered state-transition machine
model that prioritizes features and avoids interactions
due to non-determinism. A tabular notation for
specifying features has been developed. These
specifications are composed incrementally, and a
reachability graph of the composite system is
generated. This thesis demonstrates how reachability
analysis has been used to automatically detect six
types of feature interactions, with an emphasis on
telephony features.
Unix Tutorial Part II
by
Unix Shell Basics
Now that you are accepted among your peers, able to send mail and read
news all around the world, you want to get more from your Unix account. Let
usshow you how to get Unix to do more work for you!
This tutorial is intended for users with some experience with Unix and
will be held in an X-terminal lab so that participants can immediately
experiment with the material being discussed. Upon your completion of this
tutorial, you will be able to use files, directories, redirection, pipes,
and apply filters. Also, some of the advanced command-line features of tcsh
will be revealed.
Math students should be sure to have their accounts activated in MC3011,
and temporary accounts will be available to non-math students who are CSC
members.
Tuesday, September 26, 1995 and Thursday, September 28, 1995
6:30-8:30 pm
MC 3006
All are welcome.
oo$$$$$ooo o$$$$$oo$$$$$oo
o$$"" $$$" o$$"" $$$$$$ ""$$o C O M P U T E R
o$$ $$" $$" o$" "$$ $$o
$$ $$ $$o $$ $$o $$ S C I E N C E
$$o "$$ $$ o$$ $$ $$
$$o "$o $$" o$$ o$$ o$$" C L U B
"$$ooo$$$$$$ooo$$" $$$$ooo$$$
A Student Chapter of the ACM
DEPARTMENT OF COMPUTER SCIENCE
UNIVERSITY OF WATERLOO
SEMINAR ACTIVITIES
DISTRIBUTED SYSTEMS SEMINAR
-Tuesday, October 3, 1995
Celine Valot, Departement d'Informatique et de
Recherche Operationnelle, Universite de Montreal, will
speak on "Coding Partial Orders: Application to
Distributed Executions".
TIME: 10:30 a.m. - 11:30 a.m.
ROOM: DC 1304
ABSTRACT
The underlying structure of distributed executions is
the partial order induced by event occurrences. Actual
researchs focus on finding an efficient representation
of this partial order and on properties checking. An
encoding which is correct and optimal is an encoding
which has reduced memory requirements, reduced data
pre-processing time and a correct answer to the
question "does event x precede event y ?"
An encoding, to be correct, should characterize causal
relationships and concurrency between events;
correctness is granted by the knowledge of partial
order dimension and width but those values are
difficult to obtain in the context of an on the fly
computation. Analysis of the different methods shows
that it is awkward to design a general purpose
encoding; rather, the identification of various order
classes having determined properties allowing an
efficient representation is a promising direction of
research.
The optimality problem was addressed by means of
decompositions of the partially ordered set. Those
decompositions aim at minimizing the number of element
pairs, ordered within the decompositions, but
uncomparable within the partial order. Those
decompositions lead to approximations of an optimal
encoding, by getting closer to the order width, in the
general case; optimality is achieved in specific cases.
Independently of the chosen representation, the cost of
properties checking remains very high; a property should
be checked on the overall set of the possible
observations. From the chain decompositions, properties
are checked on optimal linear extensions which is a
significant observations subset. In that context, the
cost of properties checking is reduced.