Re: QT4CG-003-05: MK to consider how fn:array-filter() could be generalized to handle array predicates on the index as well as the value.

Michael Kay <mike@saxonica.com> writes:
> The following possibilities come to mind.
[…]
> (c) redefine array:filter with the signature in (a), and extend the
> rules for function coercion so that if an arity-1 function is supplied
> where an arity-2 function is expected, the supplied value for the
> second argument is dropped. This is in fact already proposed at
> https://qt4cg.org/specifications/xquery-40/xpath-40-diff.html#id-function-coercion
> - see bullet 2.
>
> I propose we go with (c).

That sounds good to me.

> Note that the filter function for arrays (and presumably for
> sequences) will have arguments ($value, $index), whereas the proposal
> for map:filter is for the arguments to be ($key, value). It might be
> more considered more consistent to make the map:filter callback expect
> ($value, $key), but my preference is to go with ($key, $value) because
> this is the order in map:for-each().

It’s impossible to make all of the signatures consistent, so users will
have to learn where the sharp edges are.

I don’t feel strongly about this, but let me make a brief argument for
the other order. I think it’s more likely that I’ll think of
array:filter and map:filter as “doing the same thing”, so I think I’ll
make fewer mistakes if the map:filter order is ($value, $key),
consistent with array:filter.

                                        Be seeing you,
                                          norm

--
Norm Tovey-Walsh
Saxonica