OK, in response to your message: I certainly must abandon some assumptions about OIDs! I do need persistent unique handles for Things so I can consistently refer to them in my Triples and Theories.
Keeping things unique by the data itself is an interesting minimalistic proposal... apparently even if OIDs are changing, compound objects like Triples are stored will still be stored "safely"? E.g. if I store the triple [ThingA, ThingB, ThingC] this will always be (de)serialized so that it points at the right Things (even if the OID of ThingC, say, happens to change)? If this works, that would be neat. (If not, I could easily enough maintain some persistent index and use it to create handles for Things.) Your statement of what I hope to accomplish in filtering-by-theory is correct. Having Theories just be subsets of Things would be fine for a prototype. Eventually it could be nice to be able to label the elements of a subset and store such labelings; Pseudo-SQL-wise, presumably all that amounts to is another column that is ignored before the "member" function is applied a la your example. _______________________________________________ elephant-devel site list elephant-devel@common-lisp.net http://common-lisp.net/mailman/listinfo/elephant-devel
