Bob W pisze: >> The only problem are people who should appear in more than >> one "context" >> (like a workmate with whom we worked in more than one place). >> In such a >> situation we have to choose one context over another or make >> "category >> reference" (supported by some software packages) so that we >> have a link >> to one category from several contexts. > > This is precisely the problem with any hierarchical organisation > imposed on non-hierarchical subject matter, and the reason why they so > often fail. You also chose an example - genetic relationships - that > is well-suited to a tree structure, so it looks as though it would > work for all subject matter. Unfortunately for you it won't. [...] > I do try to be careful not to set these things up for subject matter > which is not really hierarchical.
Well, I agree that a _strict_ hierarchical structure does not solve all the problems with people who should appear in different contexts - although even then I would prefer hierarchical structure over a flat one and just live with the limitations (after all, in a family tree the main problem are people marrying different members of the same family - there are not many persons like this). Luckily, there is a cure for at least some of these limitations. I wrote about category references which are effectively a way for (some) databases to show one category (i.e. one person) in different contexts. They behave much like symlinks in Unix filesystems and so they solve precisely the problem I mentioned above. Pawel ____________________________________________________________________________ Domena za 90 groszy! www.nazwa.pl -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.