The <intended> datastore does not contain the template themselves

[kwatsen]

Templates are only useful (to determining the final config and validation) once fully expanded/flattened.

Since templates are expected to be expanded/flattened in (see #13), there is no need to keep the unexpanded template definitions in .

Further, having unexpanded templates in may cause validation issues, if the YANG is unable to validate them (note: this assumes that #4 is False).

[RobertPeschi]

This is maybe not an essential point, but I am struggling a bit about the rationale behind this one.
#4 says that template should be validated, which I think is a good thing (we may discuss to what extend, as https://github.com/jsterne commented on #4), so I don't really see the harm showing it in the intended as well, as a matter of orthogonality, i.e. avoiding making exceptions if not strictly needed.

(note: this assumes that #4 is False).

If a template is not validated (in the running), its config could be completely messed up, hence unusable for expansion. I guess this would be best caught in the running before any doomed expansion is even attempted in the intended.

[QiufangMa]

if I understand this requirement correctly, template YANG module is only implemented in running ds (and system), and intended won't share a common datastore schema with running. This would require a rework of conventional datastore definition in (future) NMDA-bis work, I guess.

[derajara]

To maintain the consistency of the view, that an intended is a read only representation of the running, i don't see an issue of having the template itself as part of the intended (the template being just a blueprint). I think this is also related to the #4 since the earlier we detect an inconsistency in the configuration, the quicker will be the resolution instead of waiting till the template is referred by a template-consumer