Michael Kay
Michael Kay
The datalist seems reasonably OK to me. Not sure if it's best to retain the partitioning by namespace?
The current function finder is geared towards people who know what they are looking for and want to get there as quickly as possible, which is a very different requirement...
Yes, I think you're probably right that where the default is ".", we shouldn't apply the default if () is supplied explicitly. Apart from anything else, a user-defined function can't...
It gets a bit complicated with XSLT match patterns. We have special rules for union patterns such as A|B - we split it into two patterns that can have different...
OK, I've changed PR #1442 so that `match="element(a|b)"` is now handled in the same way as `match="element(a)|element(b)"`.
Looking at the subtyping rules, and considering (a) `element(p|q) ⊆ (element(p)|element(q))` (b) `(element(p)|element(q)) ⊆ element(p|q)` Case (b) is true because both alternatives of the choice on the LHS are subtypes...
I've extended the scope of PR #1442 to cover this issue.
I think this is effectively closed by virtue of PR #1442 (which suffered some scope creep as it developed).
I'm inclined to follow the convention that the contained sequence constructor has the same effect as a select attribute. So if the template defines a select attribute, and its type...
Note, if you're going to attempt this, that I've been trying to move in the direction of making a distinction between types, and designators of types. What you see in...