Re: defining quality of OOA and OOD models

H. S. Lahman,

> Model quality is probably best measured in terms of conformance to the
> practices of an established OOA/D methodology.

sigh, might be true, though taking into account theoretical maturity of
OO, we should have hopefully by now measured how well different OOAD
methods fulfill certain quality standards rather than vice versa :-(

> An OOA model is complete when it executes correctly with respect to
> functional requirements. The exit criteria for OOD modeling is fuzzier
> because one cannot validate most nonfunctional requirements without an
> executable in hand.

might be just conventions but I like to think of OOA as something
tightly linked with RA and problem understanding - RA focusing more on
functional and nonfunctional requirements and OOA focusing more on
structural and conceptual properties of the domain system... OOA that
you are referring to - I like to think of it as OOD with all it's
different levels of abstraction, and what you refer to OOD - I refer to
it as software architecture again with many different levels of
abstraction and different "views"...

>>From your answer, I see that you use as a criteria

1) completeness, and
2) correctness

and you correlate them tightly - presumably through the verification
using some model checker that will benchmark your specification with
respect to certain specified functional criteria. These two are
probably the ultimate criteria to what I perceive as OOD, and should
definitely be method independent.

Davor

.

Relevant Pages

  • Re: OOA?
    ... Responding to Kret... ... author's methodology are much more likely to view the book in a kindly ... I would look for a book with OOA and/or OOD in the title. ...
    (comp.object)
  • Re: defining quality of OOA and OOD models
    ... and, consequently, the particular paradigm methodology. ... The exit criteria for OOD modeling is fuzzier ... functional and nonfunctional requirements and OOA focusing more on ... acid test of an OOA model is that it could be unambiguously implemented ...
    (comp.object)
  • Re: Seans methodology again.
    ... The criteria that I've ... can be demonstrated as being an artifact. ... granite is the Microplan Group. ... You haven't made your methodology clear. ...
    (talk.origins)
  • Re: The Bet
    ... In what units does one measure "surface irregularities"? ... Sean, it's *your* methodology. ... criteria. ... algorithm set up for detecting deliberate artifact. ...
    (talk.origins)
  • Re: Seans methodology again.
    ... The criteria that I've ... can be demonstrated as being an artifact. ... Beyond this, if you could find a bunch of false positives, you'd no ... test I've presented is based on a pattern that the methodology has ...
    (talk.origins)