[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: a few questions


Gerald writes:

> the UML seems to be very effective using entity relation diagrams.

   A company in this house (http://www.autronicafire.no/) have used
   entity-relation diagrams to describe relations between different units
   in a fire detection system. Then they compile the E/R and the compiler
   outputs a graphical configuration tool to be used by service personnel.
   It is very elegant. E/R is, as you also suggest, not a toy. They compile
   a new tool whenever they introduce a new componnet type in the system,
   then the same config tool is being used for all installations.

> Sequence diagrams should not be compared with data-flow diagrams. A
> diagram shows simple patterns of sequential interactions between objects.

   MSC is almost a full-blown programming language these days, see
   http://www.itu.int/itudoc/itu-t/rec/z/ I've listened to a Norwegian
   the secretary of the standardization work (Oystein Haugen). He talks
   up under the ceiling (=is very enthusiastic).

> WARNING: please, don't jump to the ceiling while you are reading this :

  Thanks for the warning, my ceiling is getting worn out.

> concurrency! You must specify or agree on what these processes (bubbles)
> mean in relation to each other. Generally, we agree that they are in
> parallel (default).

   In SASD Yourdon technique concurrency wasn't much discussed, I believe,
   so, you are right. But NOW, to me, a bubble is a PAR-bubble.

> As you see, this development is a breakthrough for the OO community -- in
> particular the UML and Real-Time UML.

   Forget it. You don't live in California! But, then, Dijkstra did get rid
   of gotos. Maybe you need a longjmp!

   Have a nice week-end!

@ Oyvind Teig
@ Kongsberg Maritime Ship Systems, Ship Control, 7005 Trondheim Norway
@ http://home.no.net/oyvteig/