[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: terminology for purple numbers

Murray Altheim wrote:
> Eric Armstrong wrote:
> [...]
> > Ontologically, speaking, though, I don't like the use
> > of "id" because "id" suggests a 1:1 mapping. There are
> > many paths to a single node, however. That many:1 mapping
> > is inconsistent with the XML notion of "id" as a unique
> > identifier.
> There's only one hierarchical path to the node using the specific
> trajectory defined in both Purple and plink. "3b10" indicates a
> unique place in the document with no ambiguity, and this is why
> both Eugene and I have agreed upon "hid". If we were to settle on
> "path" or "pid" we would leave open why *type* of path we were
> using, and the 1:1 mapping would as you say be lost.
Starting from the context of specified "document", which is in
reality a view constructed from a collection of nodes, ordered
in a hierarchical arrangement, yes, there is only one path.    (01)

In that sense, the path does constitute a 1:1 "id" for the node
-- but only if the "document id" is assumed.    (02)

So a path id would in reality consist of 
   (documentID):(path)     (03)

That combination would produce a unique ID, given that document
IDs are also unique.    (04)

At the moment, since we are still living in a document-centric
universe, I agree that paths appear to be unique. But when
we move to the next stage, where we combine existing nodes and
reuse them (for example, in the kind of "reorganized dialog"
that is necessary for actually reaching discernible conclusions
in an online discussion) then paths will no longer even appear
to be unique. They will only be unique with the context of a
given document (view).    (05)

I think your point about "type" of path is important, too. There
are Xpath kinds of paths and Augment-style paths. I'm not sure
whether it makes more sense to add a type-identifier to the path
specification, or to assume that only one kind of path specification
will be used. I suspect the former. But being lazy, I prefer the
latter.    (06)