On Thu, 10 Jul 1997, Arthur Chapman wrote:
> However, to add to my earlier email (things move fast around here)
> we have been discussing how we might handle automatically generated
> pages. Things such as those produced out of a database, and the one
> I mentioned earlier, GIS maps prepared automatically on the fly.
Hmm, to my mind the database as a whole might want to have metadata
attached to it (in the DATASET.whatever grouping that we seem to be
veering towards) but the actual dynamically produced results from searches
of the database (the result of CGI scripts for example) probably wouldn't
need to have metadata attached as people usually try to prevent robot
indexers from delving into such areas and they are very short lived. I
certainly wouldn't attach any DC *about the result set* to the output we
generate in the ROADS project as a result of doing a WHOIS++ search on a
database (the fact that the output itself is more or less DC content is
another matter).
> Also there is a need to cater for interactive applications such as
> VRML, results of Java Applets, etc.
I agree that these need resource types as they are much longer lived.
More than one person can come along and pick them up over a period of
time (assuming the scenegraph or applet or whatever isn't itself
dynamically generated!) and so it makes sense to have indexers gather
metadata on them to allow people to find them.
Tatty bye,
Jim'll
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
* I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
|