On 5/3/07, Grassel Guido (Nokia-NRC/Helsinki) <[EMAIL PROTECTED]> wrote:

Marcos, WG,

On 5/2/07 7:54 AM, "ext Marcos Caceres" <[EMAIL PROTECTED]> wrote:

> As an aside, you raise a point that we seem to be continuously
> encountering: that is, what exactly is a widget? In technical terms,
> the widget specification will probably define a widget as a packaged
> collection of files that may include an optional manifest file capable
> of being extracted and instantiated on a widget engine. The definition
> itself does not speak of functionality/experience afforded by a
> widget. I guess widget engine is any software capable of extracting,
> processing, and presenting a conforming widget package. We don't
> expect our definition of a widget to match a user's experience or
> definition of a widget.

Could we limit what a Widget it by saying:
- A Widget always has a user interface.

I would be reluctant to say that a widget "always" has a user
interface. I can envision widgets being run as services that contact
web services based on system events but don't necessarily have a user
interface (although the widget engine will usually provide a UI to
start and close a widget). Eg: an intranet widget that monitors CPU
usage on a computer and contacts a hard-coded URL if the CPU usage
goes over a certain threshold.
The bootstrapped file might just contain:

<!doctype html>
<script src="Resources/cpumonitor.js" type="text/javascript"></script>

I would be fine with saying that widgets *generally* have a user
interface, or something to that effect.

- The user interface is defined by means of some markup documents that can
dynamically be modified by a scripting language. -- I wonder if the TAG has
created some prose that is language independent and could be referred to
from the definition of  what is a Widget.

I am unaware of such text, but I'll try to take a look. Anyone else know?

--
Marcos Caceres
http://datadriven.com.au

Reply via email to