Stu writes:
| my oversimplification is:
|
| The Dublin Core represents one attempt at what Tom Baker describes as
| 'pidgin metadata' (the bee interoperability metaphor also works). It
| purports to approach a simple core which is perhaps necessary (or at
| least useful), but certainly not sufficient.
Right.
| The path to sufficiency will require some unspecified number of
| additional 'packages' of metadata, and if you don't have a formal spec
| for parsing them *AND* understanding the semantics therein, you prolly
| don't want to look in the first place, so lets hope there aren't TOO
| many of them.
Right, you need both syntax and semantics for metadata that's more
complex than DC, and even DC can refer to external stuff (scheme=).
| I suspect this is not radically different than what Terry is asking
| for? If the criticism is that we have chosen
| yet-another-way-to-skin-the-cat, I'm not sure I can assert otherwise.
Sort of. We have adequate syntaxes, I think, and MIME M/R is one of
them. I'm also beginning to see that MIME engines are going to be
very widely deployed, so (while I kinda don't care for MIME) I'm
beginning to think about solving problems by using them. If M/R
will work, then the syntax part of WF is indeed just another way
to skin the cat (if we can still use that metaphor in polite company ...).
(You could call this the semantics of syntax as opposed to the
semantics of content, but let's not: we'd never dig ourselves out.)
I'm not entirely sure that any overarching syntax is required.
DC defines the syntax and semantics of scheme=, after all, and whatever
the scheme is, it defines its own syntax and semantics, including the
syntax of reference to other things.
Ron said he was looking at using M/R's syntax for a "start" piece:
> M/R allows one body part
to be denoted as the "start" of the multipart message, by default
it is the first bodypart in the message.
As in the case of SGML, if you know where to start (and one or two
other items), the whole assembles itself as you go along. So I think
Ron's approach would work, and calling it an implementaion of WF
packages doesn't appear (to me) to add anything. One does need to
say that a particular M/R message contains metadata, and perhaps
even that the metadata is DC, or MARC, or whatever those museum
informaticians talk. But M/R gives you the package concept already.
I'll reply to Jon separately (tell me if I'm just being argumentative
and I'll shut up).
Regards,
--
Terry Allen O'Reilly & Associates, Inc. [log in to unmask]
"In going on with these experiments, how many pretty systems do we build,
which we soon find ourselves obliged to destroy?" - Benjamin Franklin
A Davenport Group sponsor http://www.ora.com/davenport/
|