search

ietf-ccamp-wg / ietf-network-inventory

3 stars 5 forks source link

Questioning explicit modelling of room. #86

Open nigel-r-davis opened 1 year ago

nigel-r-davis commented 1 year ago

https://github.com/ietf-ccamp-wg/ietf-network-inventory/blob/e633109c5dff36ea26e1844c6abd32e664eb9f0a/draft-ietf-ccamp-network-inventory-yang.txt#L423

It does not seem appropriate to model rooms explicitly (room etc. structures are complex and vary from case to case). It would seem more appropriate to provide a generalized address structure to allow the non-discoverable location information to be recorded against the device. This would be of the form of an extensible enumeration (using identity) where the base model would provide building, room, floor, cabinet, hut etc. as candidate labels. Alternatively this could just be a free format string or NVP List. The NE should be the start of the modelled structure (as per TAPI Device). I have a slight preference for the term "Device".

sergiobelotti commented 1 year ago

Not totally clear the message here: equipment-room has already a "location" attribute that is a string that would be recommended to contain a GIS address. What would be the differecne with respect your proposal?

nigel-r-davis commented 1 year ago

Meeting 20230621: Agreed that there is an issue here. Agreed that the current YANG may be both too specific and not complete. A simple string was considered too open and ambiguous (not validatable).

Location in device etc. could be geographical, room based etc. May need location list where each location is a list of perhaps stings or enumerated strings where the enumeration is ROOM, BUILDING etc.