From: David Z. <dav...@gm...> - 2005-04-29 05:35:04
|
FeatureAttributeType extending ListAttributeType with the constraints specified sounds like a good compromise. It also paints a consistent road for the future should we want to revisit this in the future. David On 4/28/05, Jody Garnett <jga...@re...> wrote: > David Zwiers wrote: >=20 > >I recall it had something to do with multiplicity, but no in principle > >it should be fine (note this adds a few method to FT). There may have > >also been an issue with creating values again too (for example > >defaults). > > > Understood :-) The one reason I asked rather than recommended... >=20 > >One of my version did have FeatureType implementing > >AttributeType. I liked the result, but ended up with duplicate methods > >(and some inconsistencies with general xpath usage). > > > > > We already have those if you think about the use that *excepts* clause > in the origional FeatureType. > There is so much overlap between FeatureType and ListAttributeType that > something really should give. >=20 > >Oh, and no it wasn't flattened ;) (hence some ofthe issues with > >creation) ... flattening this would be very confusing to complex > >feature users. > > > Well I now know how Chris felt; stuck between what is needed to go > forward, and what is required for backwards compatability. >=20 > The FeatureAttributeType does not look too bad and paints a consistent > picture with respect to schema. > I am willing to go with it unless you want to think about this some more. >=20 > Another option is to make it a ListAttributeType <--- > FeatureAttributeType, and demand that all the list methods > delegate to the FeatureType. >=20 > Right now I am trying to get > FeatureTypeBuilder/FeatureTypes/FeatureTypeFactory/DataUtilities under > some kind of control. >=20 > Jody >=20 > |