[xml-dev] Process analysis vs. data analysis (was: XML and OOP -- can this marriage be saved?)

[Jeff Lowery <jlowery@scenicsoft.com>]

Here's a tangent:

Can data analysis be pursued independently of process analyis?

I'm of the mind that once you've scoped the problem (not that that ever
really happens), you can work on the data model and process models
independently, albeit with frequent and necessary sync-ups to make sure the
that two models are compatible.  The two models, separately developed,
provide a means of validating both groups' understanding of the problem to
be solved.

I think top-down (OOP) anaylsis dismisses too readily the underlying,
pre-existing facts on the ground: reports, database schemas, directory
structures, file content, message contents, entry forms, physical objects,
etc.  All these are all in place and available before any new work takes
place. You can't process data that isn't already "there" in one format or
other.

I see API's as process-think,  schemas as data-think.  The two have to mesh,
certainly.  But data is more readily determined, and more stable, than
processes that involve higher-level (and subjective) conceptual thinking.
Analysis should come up from the bottom (data) _and_ down from the top
(API). And care should be take that both models meet in the middle,
certainly.