[
Lists Home |
Date Index |
Thread Index
]
"J.Pietschmann" wrote:
> I'm always quite surprised why people insist on a more detailed structure
> with implied semantics, because the structure proved much less useful in
> practice than many people think:
Profoundly wise conclusion! As always, the useful structure is that
instantiated locally on the output of an instance parse. Such structures--and
only such structures--are specifically intended for the particular demands of
local processing. Whatever structure is (necessarily) exhibited by the markup
document itself should--must!--be considered specific only to that instance
document. Every subsequent use of that document requires an autonomous process,
which in turn requires an idiosyncratic data structure as its own input. Such a
specialized data structure can be created accurately only locally to the
process which consumes it.
The processing of markup documents is the converse of programming to a
disclosed public interface. The structure of a markup document expresses the
particular understanding of the process which creates that document for, we
presume, purposes particular to that process. The document is created without
reference to, and probably without knowledge of, how it will be consumed, and
for what purposes, by which subsequent processes. Therefore, when Joe Chiusano
writes:
> One way is to use a context-driven mechanism at runtime that "applies" the
> proper set of elements according to the context of (in this case) country of
> origin.
he is, in fact, describing the only viable mechanism for producing each locally
required data structure, but I do not think that he understands how that
undercuts the very universal standardization of document forms which he hopes
to see promulgated by the UBL, or by any similar Babel project.
Respectfully,
Walter Perry
|
