> -----Original Message-----
> From: Philippe AMELINE [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, July 31, 2001 5:34 AM
> To: [EMAIL PROTECTED]
> Subject: Re: gnumed progress
>
>
> > One can take things to extremes (and have great fun), but I
> would say
> > that for arbitrary user-defined datatypes proividing a
> table like that
> > with a starting population derived from existing users, and with the
> > manual field pointing to a description of what entries of that tuype
> > should mean and how they should be used is quite sensible.
>
> And what about providing an ontology for the labels,
> and - instead of storing individual datas - allow storing as trees,
> and - instead of storing only (field=value) datas - allow
> genuine clinical
> description
>
Implementing tree structures in a relational table is not too difficult
as long as you have unique IDs for each entry and properly maintain a parent
field for each item. You get the bost of best worlds. .. .
> Seems you get close from Odyssee (and have great fun)
>
> Philippe
>
