Re: More dictionaries changes? (was: [Bug 12248] Make objects first-class API citizens)

Dominique Hazael-Massieux:
> A few more changes that I think are needed:
> * in 3.3.3, "The identifier of an argument is given by the identifier
> terminal in the Argument, and the type of the argument is given by the
> Type. If the Type is a scoped name, then it MUST resolve [...] to an
> interface or typedef." this should reference also dictionaries
> presumably

Yeah.  (I did a search for #dfn-resolve, but I forgot there would be
unlinked occurrences of “resolve” too.)

> * likewise, I think 3.8.15 should reference dictionaries as possible
> resolution of type names

I’m not sure what you mean here.  3.8.15 says what ScopedNames that
resolve to interfaces mean, and 3.8.16 says what ScopedNames that
resolve to dictionaries mean.

> * in 3.4, "Dictionaries MUST NOT be used as the type of an attribute,
> constant or exception field"
> - Can dictionaries be used as the type of a dictionary member?

Yes.

> - Can I use arrays of dictionaries in either of these positions?

Good question.  I think we should disallow sequences and dictionaries
from being used as the element type of an array – each array element is
accessed kind of like attributes.

> - Is it that Dictionary can only be used as part of operation arguments
> (and if so, shouldn't it be stated as such)?

I think the entire set of places they can be used should be: operation
arguments, operation return values, typedefs and dictionary members.

-- 
Cameron McCormack ≝ http://mcc.id.au/

Received on Tuesday, 31 May 2011 23:51:12 UTC