NeTEx At least the id attribute (but I would argue version too) should have use=required

When there are key identity constraints stating that id + version should form a unique combination it looks obvious to me that id and version attributes should have use=required.

It does not make sense that virtually every "ref" attribute has a required attribute "ref" but there is no required attribute "id".

Jul 11 '23 14:07 skinkie

The id is really mandatory only when it is a stand-alone object that can be referenced ... when used embedded in another object, you don't really care

Sep 05 '23 16:09 Aurige

If used in an embedded object, references can still follow to that object. QuayRef was a perfect example of that case, Quay is an embedded object. So we should define the granularity of objects. There are first, secondary and ternary objects. And I agree there is an importance where we really don't case, versus where ids really should exist.

Sep 05 '23 16:09 skinkie

Agreed on my side... needs to be approved by the group

Sep 07 '23 13:09 Aurige

Start in v2 (on any possibly first class object.. any object having an "inFrame" possibility) but postpone breaking changes in v3

Jan 17 '24 15:01 Aurige

@skinkie How do we proceed? Do we have a list of the first class objects?

Feb 02 '24 11:02 ue71603