Actually, a globally unique ID of the kind you need for peer-to-peer
interchange can get quite large. A time signature, userID, and
a globally unique computer ID are needed. It's a lot, but if memory
and disks keep following the curve they have promised, it will be
a major don't care by the time the thing is ready to get out of
the lab.
Lee Iverson wrote:
>
> In message <001301c09826$19a99fe0$2312d1d1@computer>, "N. C a r r o l l" writes
> :
> >Lee, how much storage would you anticipate metadata
> >needing in the node/path scheme?
> >
> >E.g., A DKR consisting of:
> >
> >100 straight-text HTML pages
> >of 20kb each
> >and 2kb each of added author metadata
> >with ? grains specified (lines, characters?)
> >
> >= [Kb total storage?]
>
> Well, each node has mandatory fields for (forgetting any?):
> o ID
> o changed & changedBy
> o created & createdBy (maybe only on head versions)
> o changedBelow & changedBelowBy (maybe only on structuring nodes)
> o permission
>
> Now, all of these are single-word additions to node size (either 32 bits for
> id or timestamp or a pointer to user or permission record).
> So that's 32 bytes per node, not generally too bad.
>
> >(I'm ignoring path data storage here.)
>
> Don't ignore it. Put simply, Paths are client-side, transient objects --
> they are context holders. Path Descriptions would, however, be stored in
> the Repository.
>
> -------------------------------------------------------------------------------
> Lee Iverson SRI International
> leei@ai.sri.com 333 Ravenswood Ave., Menlo Park CA 94025
> http://www.ai.sri.com/~leei/ (650) 859-3307
This archive was generated by hypermail 2.0.0 : Tue Aug 21 2001 - 17:58:02 PDT