RE: RDFS versioning

: Sean B. Palmer:
:Don't forget that the SW aim is that eventually both A and B can
:convert their systems from using "a" schema to "aa" schema
:automtically by semantic botostrapping.

Yep, that's why I said I wanted to ignore discovery ... different
architectural issue.

:O.K.... let's say A uses "a", and B uses "a". Now, C comes along with
:"aa". A tries to read something of C's (written in aa), but can't.
:Luckily either the aa schema has assertions about how it relates to
:"a", allowing A and B to upgrade, or these assertions are found in a
:third party index (ahem...SWAG...ahem).

This is the hard bit. I call it metalinking. Very long links imply issues
with semantic drift. I think this will solved out of band by people hand
tuning stuff for some time to come ... no bad thing really.

:Versioning is good... but I thought that part of the SW was how to get
:around versioning?

Couldn't say, automation seems to the main goal.

Bill de hOra

Received on Thursday, 22 February 2001 10:39:27 UTC