You will have to excuse my failing eyesight - I read 'legal_entity' not 'legal_identity'. I realise now that Sebastian was referring to the has_legal_identity() function in the invariants of ACTOR. I think there is a problem in the model here - we either should get rid of this function, as Sergio suggests, or else we need to add something more to the model, so that the function can indeed be impelmented in a reasonable way. Which we do depends on whether we think it is reasonable to force all ACTORs (in all openEHR demographic data, forever!) to have a 'level identity'. This is the debate we need to have. My feeling today is that it is probably too restrictive.
- thomas On 31/08/2010 11:16, Thomas Beale wrote: > On 26/08/2010 09:27, Sebastian Iancu wrote: >> Hi Ian, >> >> I was not thinking on names attribute of an archetype as holding more >> than names. Yet, the demographic_im.pdf is suggesting/stating to use >> them to associate a type meaning the the owner objects (i.e. things >> like PERSON/name/value = "PERSON" or ROLE/name/value = "General >> practitioner"), and for some objects the purpose() is designed that >> way as well. >> >> As you previously said, there can be RM-types and Archetype-types and >> the later is introduced in demographic package through this >> construction of 'name' attribute. As long as the scope of that 'type' >> is within (or related) that owner archetype domain I don't see any >> problem, but if that that 'type' need used outside I don't really see >> it working, without a proper coding system or at least a binding. >> >> Maybe I was not very clear in previous emails about my reasoning, >> maybe I am just confused about specifications or about the modeled >> archetypes on CKM, but nevertheless one of my main technical question >> remains: >> /how can a function like ACTOR.has_legal_identity() be implemented >> regardless the archetypes being used? / > > that can only be done if the concept 'legal_entity' is also hardwired > into the information model, which is exactly what we are avoiding via > the use of archetypes. A function like this would make more sense with > a template-generated programming object, not the base RM objects. A > template, based on specific archetypes might have a legal_entity > defined somewhere, and this could be queried with a hardwired function. > > Otherwise the approach has to be to implement functions like this > using AQL or similar queries that make use of paths from available > archetypes. But in this case you would not be doing > ACTOR.has_legal_entity, but a query over the whole demographic > repository, or some subset of it, which searches for ACTOR objects > containing the specific path. > > - thomas beale*** > * -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openehr.org/mailman/private/openehr-technical_lists.openehr.org/attachments/20100831/c127d4e5/attachment.html>
