Re: two nasty schemata, union types and surrogate keys

> So in the end, we'd really like a total invertible function between the entities and their representations in the database. <

This wish is quite understandable, but I do not see one for the situation you described. There are a lot of difficulties as already mentioned in previous posts:

• we deal primarily with assumptions
• the GIGO problem (bad or incomplete sources, human input errors)
• lifeforms are not encoded in a database compatible way when they appear on earth
• social data is not easy to record as life bears lots of surprises in it

Some horror stories can be found if you look at the genealogical community. This is even more difficult as you deal with persons which are not alive anymore and with old and contratictory source documents. In the end each database assigns some magical auto-number or GUID to each person and if you try to merge two databases you find yourself in an unpleasant battle with the duplicates.

What however can be done is, that the one who maintains the internal encoding scheme is doing it with utmost care in order to minimize incorrect assignment of details. For the rest there is the UPDATE and DELETE command.

brgds

Philipp Post Received on Wed Sep 23 2009 - 11:19:14 CEST