| Date: | Wednesday, September 23rd, 1998 |
| Time: | 7.00pm-9.00pm |
| Location: | Xerox Palo Alto Research Center George Pake Auditorium |
| see directions below |
MTVJUG completes 36 months coming December'98. MTVJUG was founded in December '95.
| Agenda: | ||
|---|---|---|
| 7.00 to 7.05 | - | Announcements |
| 7.05 to 8.15 | - | Removing the "Black Magic" from Java Multithreading |
| by Paul Peterson | ||
| 8.15 to 8.30 | - | Multithreaded Debugging Demo |
| 8.30 to 8.45 | - | Q&A |
| 8.45 to 9.00 | - | Product Demos |
| (All are welcome, no membership fee, no prior reservation necessary) |
| Note: For the "Announcements" section if you have something you would like announced, or any "News" (upcoming conferences, information about similar groups, applets...etc) please send mail to (Sudhakar Ramakrishnan) sudha@best.com prior to the meeting. A bulletin board would be placed for product announcements/job openings/miscellaneous announcements. |
by Paul Peterson
The current state of developing custom multithreaded code in Java is one part software engineering, and one part black magic.
The software engineering part is learning the design patterns that control how threads are allowed to communicate. Reducing the places where threads can access shared variables allows a good programmer to write a correct program.
The black magic part starts when you think you have applied the correct design patterns, but the program occasionally fails, usually at the customer site. Threading problems are sometimes portability problems due to the definition (or lack of definition) of threads in Java. More often the problem is caused by incorrect usage of synchronization.
The talk with cover some design patterns which allow correct threaded programs to be written, some examples of incorrect programming, and a discussion of a new dynamic analysis technology that removes the need for black magic by pinpointing the location and cause of the threading problems.
Paul Petersen received a PhD from the University of Illinois working in the area of high-performance parallel computing. Currently working at Kuck & Associates, Inc (KAI) based in Champaign, IL, Paul leads the development team producing analysis tools for parallel and threaded computing.
Xerox Palo Alto Research Center
3333 Coyote Hill Rd
Palo Alto, CA 94304
(415) 812-4000
>From 280 south
>From 101