Michael Kay

>Are there compelling arguments that speak against the initial suggestion? I think you would have to define the semantics (as XSLT and XQuery currently do) in terms of bottom-up construction...

>Wouldn’t it be possible to move the existing node constructors to the XPath specification? Very reluctant to do that. It's a horrible spec with its crazy rules on namespaces, whitespace,...

Let's put this on hold for a while. I want to explore implementation options. But let's do JNodes first.

Would you want a function that handles any map, or only the kind of map that might have been produced by the element-to-map function? I think it would be good...

This raises questions about streamability of context-dependent functions. Clearly we can't allow the captured context of a function like current-group#0 to contain a reference to a streamed node. This is...

I'm inclined to the view that the caller of doc-available shouldn't have to worry about different reasons for non-availability. Better to proceed with the open issue that asks for an...

I think there may be a case here for a two-argument form of `get()`, so given `{"a": ([1,2,3], [4,5,6])}`, `/a/get(2, 1)` should return 4.

My assumption is that get(x, y) would be (broadly) equivalent to a predicate `[jnode-position()=x and jnode-selector()=y]` (or vice-versa).

Given `{'a': 23, 'b': ([1,2,3], [4,5,6])}` `./child::b` returns a single JNode with content `([1,2,3], [4,5,6])`, and `./child::b/child::*` returns the flattened sequence (1, 2, 3, 4, 5, 6). I do think...

@ChristianGruen Yes, I think that's roughly what I had in mind. I'm not sure about the name `content`. I was thinking perhaps `sequence` to emphasise that you only need this...