Proposal to support optional parameter values on static functions.

[rhdunn]

This proposal allows users to define optional parameters for a function. When calling the function statically (e.g. f(1,2,3)), or creating a named function reference to it (e.g. f#3), any optional parameters that are not specified take the default values.

Motivation

Defining functions that can take one or more optional argument (such as in the Functions and Operators specification) adds a lot of boilerplate code. For example, `fn:tokenize could be written as:

declare function fn:tokenize($value as xs:string?) as xs:string* {
   fn:tokenize($value, "\s+")
};

declare function fn:tokenize($value as xs:string?, $pattern as xs:string) as xs:string* {
   fn:tokenize($value, $pattern, "")
};

declare function fn:tokenize($value as xs:string?, $pattern as xs:string, $flags as xs:string) as xs:string* {
   ...
};

In this proposal, the above can be written more simply and concisely as:

declare function fn:tokenize(
   $value as xs:string?,
   $pattern as xs:string := "\s+",
   $flags as xs:string := ""
) as xs:string* {
   ...
};

Both of these definitions of fn:tokenize are equivalent.

Note

Once this proposal has been accepted, we should go through the XSLT and XQFO specifications and use defaults for all the relevant functions.

[michaelhkay]

This is a good start, but the proposal is far from complete. Giving an example of old syntax and an example of new syntax and stating that they are equivalent does not constitute a specification. In addition, defining the semantics of the default value using this approach has problems. It would prevent the use of a default such as ".", because "." is undefined within a function body. If we want to allow that, then specifying the semantics of a variadic function declaration in terms of a lexical expansion to a set of non-variadic function declarations isn't going to work.

There's a lot of prose in the spec that presumes a one-to-one correspondence between function declarations and functions in the static context, and therefore a lot of editorial work to handle the fact that this can no longer be assumed.

[rhdunn]

I wanted to specifically only define the problem initially, not go into the specifics. That is, this is a high-level view of what is to be achieved. Think of this draft as being from the user's perspective.

[michaelhkay]

A pull request to implement this for XSLT and XQuery has been submitted.

[michaelhkay]

I believe this can now be closed as completed.