Uche Ogbuji wrote:
> Joe, BTW, I'm with you on selecting #2. #1 and #3 probably mean that at some
> point there will come a need to break down the semantics in the complex GI, and I
> think that GIs should be as close to atomic as possible for the expected life of
> the data.
BTW my personal prejudice is also for #2, *but* #2 is just as likely to need to be
*composed* into a more monolithic form as #1 and #3 are likely to need
de-composing. And that's just the point. Once we posit the internetwork providing a
web relationship among processing nodes, once we truly expect reuse and
repurposing, it makes no sense to chose a markup style based on an anticipated use
downstream: the most valuable uses over time will be the unanticipated ones.
Respectfully,
Walter Perry
