On 5/06/2009 5:27 PM, Francis GAYREL wrote: > To build a consistent oriented tree we need to associate to the nodes a > ranking property such as the birthdate (or any precedence criterion). > Therefore the ancestor of someone is to be selected among older ones.
"Ancestor" is a *derived* relationship, not something you'd wish to store in the database, and is quite irrelevant at data-entry time. Don't you mean "parent"? Let me get this straight: the user is entering the details of animal X who was born yesterday and has to input somehow the identity of the mother and of the father. > To make the ancestor allocation more easy the ancestor's list may be > filtered on birthdate credibility. So your plot is, (e.g. for the father) to show a list of all male animals who are in some credible-parenthood age range on (say) a drop-down list, and the user selects one, hopefully not at random. Is that right? I would imagine in a planned targeted organised animal breeding program that the mother-to-be and father-to-be are recorded at the time of impregnation, and the identities are established from ear-tags, embedded chips, photographs, etc and the credible-parenthood test is applied then [note: test, NOT input selection method] and all of the above is confirmed at birth. > The ranking property eliminates the circular link concern. Indeed, but you have to use it properly to eliminate other data integrity concerns :-) HTH, John _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users