From: Martin S. <mar...@ma...> - 2012-09-20 13:33:45
|
On Thu, Sep 20, 2012 at 09:47:37AM +0200, Benny Malengier wrote: >2012/9/20 Martin Steer <mar...@ma...> > > > According to Wikipedia, "In the GEDCOM lineage-linked data model, all > data are structured to reflect the believed reality, that is, actual (or > hypothesized) nuclear families and individuals." Gramps, at least in its > UI, has inherited this attachment to the nuclear family. > > If I were rolling my own, I wouldn't use a family construct at all. I'd > do something like: > > parent > | > v > birth event <- child > ^ > | > parent > > parent > | > v > adoption event <- child > ^ > | > parent > > and leave it at that. The event_type, together with the link, > establishes the relationship. Notice that this allows any number or kind > of parent, of any sex. Siblings have a parent in common. > > Are we there yet? > > >Yes, many people hate the family construct. It does have it's benefits from a >technical viewpoint in my view. It simplifies looping trough families to >discover in an automated way relationships. For example, the fanchart I just >finished in trunk need not figure out how things work, the family holds the >information. In your setup, a lot of extra information will have to be provided >to do a correct reconstruction. On the other hand, use of the family in Gramps complicates the interface. In my setup you have individual + event (= relationship). Gramps has individual + event, individual + relationship, and individual + family. > >So, for a more general construct, I personally would just allow familygroups to >be constructed, for those cases that a man has several women or vica versa. But >again, in my view it would be connected to living together somehow. So the >Sheik of Saudi-Arabia could have a family group, but not eg Mitterand or the >King of Belgium. > >Benny > |