Re: MultiValue Databases

> The query [relationship between John and 111-1111] was pointless, but it was instructive to show that it could be handled.

The point was, in the simple example, the xrdb solution did not require user to update schema and possibly transfer data to new db to handle new data/queries that were not known/anticipated at initial design time. Wouldn't a methodology that is less affected by new requirements, be an indicator of possibly being more general?

Also to show that RM was able to handle it, could you post the missing RM query to find a person with phone#s 111-1111 and 222-2222 using the new RM schema. (Note that xrdb's original query continues to work)

Also, how can I implement the following xrdb queries in the new RM schema. (Note that xrdb does not require any schema changes to support the following queries)

// Find like's type.
// Returns verb.
(SELECT * inst like)

// Find a person who likes a person whose phone# is 222-2222.
// Returns john.
(SELECT * like (* phone# 222-2222))

// Find type of thing
// which is liked by a person
// whose ph# are 111-1111 and 222-2222.
// Returns person.

(SELECT * inst ((person inst * & * phone# 111-1111 & * phone# 222-2222) like *)) Received on Thu Jun 09 2005 - 22:48:17 CEST