Re: PLink availability/feature requests

From: Eric Armstrong (
Date: Wed Apr 25 2001 - 14:38:11 PDT

Murray Altheim wrote:
> Eric Armstrong wrote:
> >
> > My expectation was that the combination of
> > * username
> > * system name
> > * timestamp
> >
> > is necessary for unqiue identification.
> >
> > The use of system name was necessary to avoid overlapping
> > timestamps, since two systems *could* have different clocks,
> > and a person could easily create a node from two different
> > systems.
> >
> > Unfortunately, that is a lot of overhead for every node.
> > I didn't see a better solution, at the time, but I'll
> > happily entertain a better idea!
> If you specify this information on a per-document basis rather than
> a per-node basis and let the document act as the "namespace" you'll
> not be adding much overhead at all.
Tempting. But if document is a considered as a "view" that combines
a collection of nodes into a hierarchy, it means that a node can
exist in multiple documents.

To inherit information from the "document", it is necessary to
accord priority of place to the *first* document the node was
constructed in, and to never-ever delete that document.

Otherwise, you wind up with nodes that have lost their original
parent, but which still exist because they have been incorporated
into other documents. And in that case, from whence can a node
inherit such information?

Perhaps a "system/author header" would make more sense to catch
such info? The header would then list any and all nodes created
by the author while on that system, and remain in place until
the list was empty (assuming that node deletes are supported).

This archive was generated by hypermail 2.0.0 : Tue Aug 21 2001 - 17:58:05 PDT