Friday morning at TRI-Ada 96 started with an unusual, but nice, occurrence;
there was a waiting line to get into the opening session. It was nice to see an
8:00am crowd big enough to cause a little traffic jam at the meeting room.
Next year s TRI-Ada 97 will be November 9-13, 1997, in St. Louis, Missouri.
They ll be looking for organizational help and papers. Start planning now. I
also expect them to post a lot of information on this list.
Watts Humphrey (SEI) gave the Friday opening keynote, "What if Your Life
Depended on Software?" The main discussion was about defects in software. Even
if the defect rate is very low (percentage wise) the number of defects in a
large program is large. Most software engineering technologies are oriented
toward finding or preventing large design errors, but little mistakes (wrong
initialization, typographical errors, etc.) can still slip through. These
"little" errors can still have a tremendous impact.
He talked about the capability maturity model and how he has related that to a
personal software process. His approach to software quality is to put more
emphasis on writing programs correctly from the beginning. Improving measurable
quality requires measuring quality (defects per thousand or million lines of
code). He showed data where it would have been cheaper to start over than to try
to test in quality. He also talked about their training in the personal software
The second Friday keynote was by Robert Dewar, "Integrating Legacy Components
with Ada." He talked about interfacing Ada95 to other languages in a portable
way. There is a spectrum of coordination ranging from the formal standard to ISO
WG9 ARG (formal language interpretations) to the ARA ACE (Ada Resource
Association, Ada Common Environment) effort to informal coordination.
Later I was drafted to be part of a panel on Ada and Java. Of course Tucker Taft
was part of the panel as was Ken Arnold (JavaSoft). The panel stressed the
advantages and similarities of the two languages. ISO/IEC JTC1/SC22 has formed a
Java Study Group. Their mailing list can be joined by sending a subscription
message to [log in to unmask]
I hope others will report on other parts of the program.
- Bob Mathis