Lists Home |
Date Index |
> Yup, this is a pretty fundamental divide, but I think it's more than a
> "personal thing" There was a thread (well, Leigh Dodds and I corresponded
> a bit) at http://lists.xml.org/archives/xml-dev/200112/msg00778.html and
> http://lists.xml.org/archives/xml-dev/200201/msg00006.html trying
> to define
> decision criteria for when a "strong" (up-front design, strongly
> typed, tightly
> coupled) or "loose" (build it simple and let it evolve, untyped,
> loosely coupled)
> approach makes the most sense.
I remember the original discussion and remember thinking at the time that it
provided some valuable insights. One other consideration that just came to
me was knowledge of the problem domain.
Well understood domain -> strong
Poorly understood domain -> tight
If a company is building its 15th X and has 'experts' who deeply understand
the problem domain of X, then there are benefits that can be achieved from
an 'engineered' approach. I don't know if Al works in this sort of an
environment. But this is certainly an environment where an 'engineered'
approach can be better suited to a rapidly changing requirements, etc.
In contrast if the company is building the first Y in the world, then it
might find an XP approach much more suitable.