Re: [Gnumed-devel] LDAP vs. SQL - RT comments

[Karsten Hilbert]

Richard wrote:
> Again I come at contacts from a pragmatic point of view, not a technical one.
Good.

I think I understand your point. You seem to be saying that
for fast and efficient target contact selection we need to be
able to control the contacts schema directly. Therefore I would
like to propose this solution:

Have an XML-RPC server "contacts". This server spans possibly
several sources: a) The service "administration" which includes
the practice's configuration, eg. the practices contacts
database. This is the schema we control and use for
efficient/rapid selection. b) The service "reference". This
will contain reference data for those countries where "official"
contacts data is distributed on CD or by download (eg. in
Germany). We also control this schema but it is also more
closely modelled to the reference source data layout. Of
course, there would be facilities to say something like "move
all cardiologists in the SF bay area into the practice contact
database" (eg filling the local contacts from the reference
source) - this may go as far as making their *content*
identical (but not necessarily their layout). c) Access to
online, remote, official repositories (these are most likely
going to be LDAP or PKI key server type things).

Again, the server "contacts" would encapsulate this into one
API. When the user hits "widen search" it adds the reference
data sources to the match scope. Any selected contact will be
transferred from the reference source to the local contacts
repository automatically (the abovementioned scripts "move all
of ... into the local repository" will still be there for bulk
transfers/updates).

This IMHO is a good way of dealing with contacts.

Other suggestions ? Takers ?

Karsten
-- 
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346