129: Context item generalized to context value

[michaelhkay]

This is a first cut proposal to generalize the context item to a context value, allowing (for example) array predicates.

The proposal covers XPath and XQuery only at this stage; it doesn't address the consequences for XSLT.

Careful review requested!

Addresses issue #129 and issue #367.

[ChristianGruen]

Very appreciated!

I didn’t find anything dubious, and it’s pleasant to see that the context item and position can also be utilized for context values (when using the new array predicate). It’s mostly the path expressions that I believe need more discussion:

1. After more consideration, I think we should discard the simple map operator, go further and treat / as ~/(fn:root(.) treat as document-node()). Path expressions are known to return ordered and duplicate-free results. A context value may contain all kinds of nodes (not only document nodes). As an example, it would be surprising if a path expression on an input sequence of nodes (that all belong to the same document) would return multiple results:

document { <xml><a/><a/></xml> }//a => { /xml }

2. If possible anyhow, we should also adjust relative paths to take advantage of the context value: a and ~/a should be equivalent (by analogy with a and ./a, which are equivalent for single context nodes). It would be great – and certainly consistent and orthogonal – if we managed to get all the following expressions working:

Items	Sequences
declare context value := doc('x.xml'); //item/name	declare context value := doc('x.xml')//item; name
$nodes -> { a/b/c }	$nodes => { a/b/c }
$nodes[a/b/c]	$array?[a/b/c]

[Arithmeticus]

Unclear what examples are meant in the preceding paragraph. The term "all" is ambiguous (object of "above" or modifier of "examples"). Better?: "In each of the preceding situations, a sequence...is processed one item at a time." Consider changing other cases of "example" in this paragraph to "situation"

[ChristianGruen]

For those who have time to check out this topic:

I have reworked the initial comment of #129 by adding explanatory comments and aligning it with the features from the latest version of the specs.

[michaelhkay]

Replaced by a new PR

[michaelhkay]

Overtaken by PR #970