John,
You're absolutely right: The raw analysis data give us 47 points of
interest in the schemas with regard to name usage. These points of
interest range from mere curiosities, to easy-to-fix inconsistencies, to
areas where deeper thinking might be warranted. Given scheduling and
other considerations, I expect that taking no action will be the most
judicious approach in many cases. On the other hand, there's low
hanging fruit, like removing the duplicate type definition for
<Proposer> or uniformly capitalizing the attribute name "reportType".
As usual, it's the cases in the middle that will take the most time.
The trigger for writing the analysis program was that EML 6.0 will soon
be frozen--even simple housekeeping chores will be difficult after
approval. Given that EML comprises a thousand some odd definitions, I
wanted some way to characterize usability aspects of the schemas as a
whole. The schemas are already in very good condition, maybe this
analysis can provide some fine tuning.
As far as moving forward goes, we could categorize the points of
interest according to several criteria, such as implementation effort,
expected short-term improvement, expected long-term improvement, and
level of disruption to installed base. Once we perform this vetting, we
can determine what's feasible within whatever time/resource constraints
we have.
Rich
