- From: Cameron McCormack <cam@mcc.id.au>
- Date: Thu, 01 Sep 2011 16:57:00 +1200
- To: Allen Wirfs-Brock <allen@wirfs-brock.com>
- CC: Andreas Gal <gal@mozilla.com>, Robin Berjon <robin@berjon.com>, public-script-coord@w3.org
On 30/08/11 10:56 AM, Allen Wirfs-Brock wrote: > And W3C could provide guidelines for when and how to introduce new > global names and how to introduce object-based name-spacing to minimize > the number of new global names introduced by emerging Web APIs. This > could be done now and doesn't have to wait for ES-Harmony modules. That is true. (For the "object-based name-spacing" I assume you mean something like hanging APIs off window.navigator instead of directly off window?) > A interface definition language specification is a very different thing > from a API design guideline document. One tells you want is expressible, > the other tells you while forms of expression an API designer should > choose. For example Microsoft .NET has a published set of guidelines > (http://msdn.microsoft.com/en-us/library/ms229042.aspx ) providing > guidance for designing the APIs of .NET frameworks and class libraries. > > Ideally, there would be a W3C Web API Design Guidelines document that > would be used in conjunction with the WebIDL spec. I agree. For the moment (since writing up such a document is a moderate undertaking), I think Web IDL would benefit from a statement about when to use [NoInterfaceObject]. I agree with Andreas that, for "concrete" interfaces, it should not be used unless there are particular web compatibility reasons not to. We should otherwise consistently expose interface objects for "concrete" interfaces.
Received on Thursday, 1 September 2011 04:57:43 UTC