- From: Jonas Sicking <jonas@sicking.cc>
- Date: Tue, 07 Mar 2006 00:41:18 -0800
- To: Maciej Stachowiak <mjs@apple.com>
- Cc: Ian Davis <Ian.Davis@talis.com>, public-webapi@w3.org
>> Maciej's initial draft [1] introduces a new TimerListener interface >> that is passed to the setTimeout and setInterval methods. This seems >> like a break to existing behaviour which is to pass a function or a >> string to be eval'ed. There has been some discussion on the lists >> about this design but no clear consensus. Clearly it's a goal to >> rationalise the window interface so that it fully supports DOM events >> but we also need to document existing behaviour. >> >> My proposal is to keep setInterval/setTimer in their two argument >> function forms, i.e. they expect a function as first argument and a >> delay in milliseconds for the second. The parameter list form has been >> shown to be non-interoperable [2]. We don't endorse support for the >> string eval form. >> >> We then define new events for the interval and timer - although I'm >> not sure about the mechanics at this stage. Should it be >> setIntervalListener? >> > My intent was doing this was that ECMAScript bindings would define a > special rule so that only functions or strings are actually accepted, > much the way functions are accepted by EventTarget.addEventListener. The > reason for this is that the "function or string" interface doesn't > really work for non-ECMAScript languages. In particular, for Java, "eva > this string" is not a sensible interface. Personally, I wouldn't mind seing support for passing a string to setTimeout deprecated. Unless someone has a good usecase of course. If we really do want to allow a string to be passed then I like the approach of defining special ECMAScript bindings to handle it. / Jonas
Received on Tuesday, 7 March 2006 08:41:26 UTC