Re: Keyword arguments and fn:serialize [was fn:slice]

On Mon, Dec 7, 2020 at 1:42 AM Michael Kay <mike@saxonica.com> wrote:

.  .  .


>
> I share your concerns here. This is a good "sanity test" for any changes
> in this area. As it happens, fn:serialize() has particular complications
> because the second argument can be either a node or a map (we introduced
> the ability to supply options using an XML structure before maps were
> available) and that makes it especially challenging; but if a new call
> mechanism doesn't work for fn:serialize then I don't think it's worth
> having.
>


@Michael Kay <mike@saxonica.com> , @Benito van der Zander
<benito@benibela.de> :  The proposed way of defining keyword
parameters/arguments (at:
https://lists.w3.org/Archives/Public/public-xslt-40/2020Dec/0101.html) does
not require any changes to the way fn:serialize() or any other existing
standard function is defined.

Nothing in it prevents any argument being a map or a record/tuple that
contains any wanted number of predefined entries.

Therefore, the statement that this " call mechanism doesn't work for
fn:serialize" is simply not true.

The proposed way of defining keyword parameters/arguments can be used with
an additional "varargs-like" mechanism (if such is really desired) for
additionally providing a potentially unlimited number of arguments when
calling a particular function.

This can be similar to the way this is described in the referenced C#
document:
https://docs.microsoft.com/en-us/dotnet/csharp/methods#parameter-arrays
<https://docs.microsoft.com/en-us/dotnet/csharp/methods>

Or, simply and preferably, have an argument of record/tuple type and define
this record/tuple in any wanted way. Why this " doesn't work for
fn:serialize"?

Please, correct your statement.

There will be coming soon a more concise and formal version of the
proposal, so please, send more questions/comments on this in order for them
to be reflected in the document.

Thanks,
Dimitre



>
> Michael Kay
> Saxonica
>
>