Minutes of meeting 22 Jan 2002

Meeting opened 1.00pm

Present: Raphael Goldberg, Peter Howarth, Daniel Justice, Michelle Osmond, Zhu Tan, Aleksandar Trifunovic

What we've done since the last meeting

• Michelle: Implemented a Petri nets prototype editor using Swing and JDOM, with a MDI. It reads in a pnml file and can, so far, draw and move places, updating the DOM in memory. Implements places on screen as a subclass of JLabel, with the real data stored in a JDOM Document object. (See Links for download).
• Daniel: Done the DOM and JDOM summary.
• Tan: Worked out OMT/UML diagrams for the conventional object approach.

Discussion

Continued discussing reasons for and against using the JDOM-based model over the usual object model. No unanimous conclusion, though most in favour of JDOM due to the potential flexibility (a new feature compared to most Petri Net projects). Decide on Friday (discussion with Will). Points raised include:

• JDOM-based:
  • extensibility. all data automatically extracted from the input xml file, regardless of whether it's part of the pnml definition we're working with now (or even might not be PNML).
  • editor could be built to be dynamic to an extent, so it could extend its own capabilities without any extra coding/compiling. (pnml may/will be extended to support different types of nets.)
  • JDOM well-supported and respected; robust code written already to deal with parsing - saves us effort, and we can trust it.
  • the more in-depth details of XML could be taken advantage of (namespaces, validation).
  • modules could also modify the JDOM if they're passed a reference to the Document.
  • BUT: need to have a separate holder (class hierarchy) to store the functionality (like drawing objects), the data is not protected, and it doesn't follow the 'conventional' methods/models we're familiar with already. How to do animation?
• Object model:
  • tried and tested (as in previous projects), the way of doing things that we've been taught - can store data and functionality together.
  • full control over the data access.
  • easier to see how to program things like the animation.
  • BUT: extending the program to support new types of net, and elements of XML, is much harder (needs coding/compiling, probably modification of large parts of the program). also would probably be passing the JDOM Document to modules anyway (no need for data protection, but modification of the active data by the modules wouldn't be possible).

To investigate:

• Daniel: Code up a (non-active) GUI to show what we discussed in the previous meeting, and use as a basis later.
• Michelle: Continue with the Petri net editor prototype.
• Tan: Try implementing the Object based idea.

Next meeting: Friday (25th), 12:30pm with wjk and then continue meeting afterwards.

Meeting closed 1:55pm