Michael Kay
Michael Kay
Caveat: this would still mean there is no defined ordering for durations that are not instances of `yearMonthD` or `dayTimeD`. We could fill that gap by sorting them first on...
While we're doing this, we should reconsider how to organise the specification. It might be best to organise all the rules under the fn:compare function, and drop the op:XXX-equal and...
We could also consider changing min/max/highest/lowest to treat untypedAtomic values as `xs:decimal` rather than `xs:double` if they are castable as `xs:decimal`, in the same way as we (are proposing to)...
>The overall question may be what we want to improve (most of all): The user experience; the structure of the specification; or the burden for the implementer? At present I'm...
The attached PDF gives a summary of the functions (&c) that make use of atomic comparisons, and the way in which the atomic comparisons are defined. My inclination is to...
It would be possible to make the second argument a function that selects the nodes to be updated relative to the root, but I'm not sure that buys you much....
Fix #77
I think the situation is analagous to the situation with attributes in XML. It's probably a bad idea to do `element/@*[3]` but we allow it and we define constraints on...
Note also that although the JSON specs say that order of properties is immaterial, many programming languages choose to retain the order, so XPath 4.0 is not out of line...
I'm marking this a "Propose close with no action" because the only way we get to put an issue on the agenda is either to raise a PR or to...