In the EBNF, use explicit separator syntax

[michaelhkay]

We could pick up the ++ and ** operators from Invisible XML for defining "a sequence of [one|zero] or more Xs separated by Ys".

So for example

Annotation ::= "%"  EQName ("("  AnnotationValue ("," AnnotationValue)* ")")?

changes to

Annotation ::= "%"  EQName ("("  (AnnotationValue ++ ",") ")")?

and

TypedFunctionTest ::= ("function" | "fn") "(" (SequenceType ("," SequenceType)*)? ")" "as" SequenceType

becomes

TypedFunctionTest ::= ("function" | "fn") "(" (SequenceType ** ",") ")" "as" SequenceType

Note that there is an impact on anyone who uses the machine-readable XML version of our grammar. However, the current version could easily be generated by preprocessing.

[michaelhkay]

I've spent half an hour looking at this and decided it's not a task to be undertaken lightly.

Apart from anything else, there's an awful lot of code both in our project and elsewhere that's driven by the grammar file, and a lot of it would break.

So I'm going to close this with no further action.

[michaelhkay]

Reopening because it turned out to be rather easier. Anyone using the grammar file, for example to generate a parser, can easily preprocess the grammar into a different form.

[johnlumley]

A preliminary test, using the XPath test suite, doesn't seem to add any additional ambiguities or errors by using the repeat/separator form.

At present the use seems to be only where the separator is always a constant string. It might be possible to extend to nonterminal or compound separators, though whether this is worthwhile is a moot point.

[michaelhkay]

For simplicity the representation of the grammar in the XML currently only allows a constant string as the separator; I made that choice deliberately as (a) it's the 99% use case, and (b) I think other cases would probably be confusing if presented this way.