Hi,

we now have two drafts with pull-style mechanism of combining data
models:

- draft-bjorklund-netmod-structural-mount-00
- draft-lhotka-netmod-ysdl-00

In many aspects they are quite similar, yet there are also substantial
differences that deserve to be discussed. In believe the underlying
problem is really pressing, and we should probably try to merge both
approaches and come up with a single solution asap.

In my view, the main differences are the following:

1. Structural mount uses a YANG extension while YSDL doesn't require any
   changes in YANG.

2. Structural mount allows for attaching sub-models only to places
   identified with "mount-point" extension, in YSDL all container,
   list, case and anydata nodes are eligible as "mount points".

3. Structural mount provides the meta-schema as regular state data, YSDL
   assumes they will be available to clients as a separate resource, or
   perhaps as an (optional) part of yang-library.

4. Structural mount puts yang-library data into the meta-schema while
   YSDL just uses references to the "standard" yang-library.

5. Structural mount allows different entries of the same list to contain
   different mounted data models (via instances of yang-library), in
   YSDL this can be achieved indirectly by defining a choice in the
   "super-schema" and assigning different schemas to its cases.

6. If the same set of modules is to be mounted in several locations,
   then in structural mount the same yang-library information has to be
   copied to all locations. YSDL offers the "schema" construct as kind
   of "grouping of modules" that can be referred to from different
   locations.

Lada

-- 
Ladislav Lhotka, CZ.NIC Labs
PGP Key ID: E74E8C0C

_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod

Reply via email to