Re: ERB call on addressing

At 08:01 PM 3/25/97 -0800, Jon Bosak wrote:
>[Gavin Nicol:]
>| >@xmlToc is the name of a script that generates the XML TOC for the
>| >book named ADVOSUG.  If the choice is between address and query, it's
>| >a query.
>| You *know* this is a script, the browser doesn't. To a browser,
>| this is an address that returns a resource.
>Then I'm not sure what "query" means any more.  How does the stuff
>generated by @xmlToc differ in principle from something a database
>might generate in response to a SQL query?  Is that an address, too?

A query is a form of address. Or rather, any address is a form of query.
All addresses are references to properties of objects held in some abstract
data structure (e.g., groves).  The only question is whether or not the
forms of address are highly restricted (such the semantics of their
interpretation are simple and easily implemented) or very general, as in
the case of unbounded database queries or scripts.

As Gavin points out, the thing receiving the result of the address doesn't
care how the result was arrived at as long as the result is intelligible.
That's why groves are so important to both HyTime and DSSSL: they define
the types of things that can be returned regardless of the form of address,
which is why in the corrected HyTime, a query location address must return
a node list, regardless of the notation used for the query or the data
notation(s) to which the query is applied.  Thus the result of resolving a
query location will always be useable to the receiver because all addresses
in HyTime resolve to node lists, regardless of the form of address.