W3C home > Mailing lists > Public > public-script-coord@w3.org > October to December 2009

Re: [WebIDL] Trying to understand IndexGetter/NameGetter

From: Jason Orendorff <jason.orendorff@gmail.com>
Date: Thu, 15 Oct 2009 14:31:59 -0500
Message-ID: <bb8868b90910151231i60c40a29ge0a6d898cc781d8b@mail.gmail.com>
To: Cameron McCormack <cam@mcc.id.au>, Boris Zbarsky <bzbarsky@mit.edu>, public-script-coord@w3.org
If I've missed discussion on the points below, I apologize. I'm just
getting up to speed on WebIDL.

1. I'm interested in contributing work to WebIDL's
getter/setter/creator/deleter features. It seems like there is general
interest in making sure this gets reconciled with ES5+, certainly
before WebIDL becomes a recommendation, but there is a fairly large
chunk of work there that hasn't started yet. Is that right?

2. I have a question about these features. The spec makes it very
clear that a Java object implementing a WebIDL interface may not
expose all the functionality. In section 3.3.4 of the editor's draft,
there is an example:

    interface Dictionary {
      readonly attribute unsigned long propertyCount;

      getter float (in DOMString propertyName);
      setter void (in DOMString propertyName, in float propertyValue);
    };

It says, "in the Java language binding, the corresponding Java
interface will have only a single method, long getPropertyCount()".

Apart from the obvious problem that Java programs can't do anything
useful with that, it makes it impossible to autogenerate wrapper code
exposing, say, a Java implementation of a WebIDL interface to
ECMAScript. Even if I have the Rhino kung fu to autogenerate
arbitrarily hairy code to make my Rhino object behave however I want,
the functionality just isn't exposed by the underlying Java object.
There's nothing to delegate to.

Cheers,
-j
Received on Thursday, 15 October 2009 19:32:32 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 May 2013 19:30:02 UTC