[comp.software-eng] SA/SD - event partitioning method

jgb@ektools.kodak.com (Joanne Green-Blose) (05/01/91)

I have a question for anyone knowledgeable in SA/SD who may have used
the event paritioning approach.  Yourdon discusses this method in his
book Modern Structured Analysis.  Basically you begin with "events" to
your system which are generated by external terminators.  The collection
of all the events to the system makes up the Environmental Model.

My problem is the implementation of this phase.  My group is attempting
to try this.  We divied up the terminators in our context diagram among
us and gave each individual the action item to find events for their
respective terminators.  Then we review the events as a group.  We
review them mainly on legitimacy (is this an external event generated
outside the system for ex.) and also to add/subtract other events.

The problems we are having are many but one of them is in the
granularity of the events.  We can't seem to be consistent on how
granular to get.  For example, a lot of our system is very
data-intensive and do we describe the event as for ex., "host computer
downloads data" or do we break it down further like,"host computer
downloads xyz data and also downloads abc data" etc.  Does it make any
difference?  Does it make any difference if some of us have generic
events while others are detailed?

Also, is it possible to have event lists for some of the terminators but
not for all?  Would this throw a wrench in the system when you start
attempting to draw DFDs?

Any input from anyone who has some experience with this (now quite
common) structured methodolgy would be most welcome!

-Joanne Greene-Blose, Eastman Kodak Co.