I reviewed previously revision -03 of this document. Most of my comments and suggestions have been taken into account, and I think the document has improved considerably.

I am still not happy with handling of XPath in the "where" query parameter:

- It is referred to sec. 6.4.1 of RFC 7950 for the definition of XPath context. However, that section doesn't define the context node, so section 3.1.1 should state that the context node is the target of the query (I assume). Moreover, the accessible tree as defined in sec. 6.4.1 of RFC 7950 depends on circumstances that are specific to YANG, so I think an explicit definition of the accessible tree should be given in sec. 3.1.1.
  
- Sec. 3.1.1 also says that prefix binding are further defined at the protocol level. However, I don't see anything like this in draft-ietf-netconf-list-pagination-nc-07.

- Standard XPath is very liberal in terms of missing nodes – they almost never lead to an error. I think a similar approach could be adopted here, too: an expression referring to a missing node could simply lead to a filter expression that is always false.

- I don't understand at all why nodes that are optional and conditional in the schema are not allowed. Note that e.g. an optional leaf may be defined with a default and then everything should be absolutely fine. This also applies to the "sort-by" query parameter.

**** Nits

- The document uses "descendant" and "descendent" interchangeably. Whilst both forms are possible, I'd suggest to stick to one of them, probably the former.