[ng-dhtml] [Dojo-interest] Dojo Update

Erik Arvidsson erik at eae.net
Tue Sep 6 17:24:26 CDT 2005


I'm more along Martin's lines here (not too surprising with my 
background). However I believe dojo has some really interesting 
possibilities here with its quick prototyping and template system. I 
just believe it needs more building blocks.

A good example of what I mean here is a custom slider. One can make it 
from scratch taking care of focus, keyboard and mouse events or one 
could build it from lower level components like repeat buttons, drag 
handles and so on. (There was a nice demo of this with Avalon at the 
MSFT PDC a long time ago.)

On the other hand one cannot expect people to reimplement grids and 
trees over and over. These are really big beasts.

The question whether to keep dojo widget free or not is a matter of 
taste. The good part in removing widgets is that dojo can concentrate on 
other stuff and the bad part is that it would require other people to do 
these and then the core might not be efficient, usable enough to do it.

In the end, what people do want is rich and flexible UI and I'm afraid 
that if no official GUI toolkit is provided with dojo, the GUI will 
become a mess like it did with DynAPI and DomAPI.

erik

Paul Sowden wrote:
> On 6 Sep 2005, at 11:28, Alex Russell wrote:
> 
>> So if we're serious about finishing this platform and exposing the  
>> new and latent browser features that will let us collectively  
>> compete with a XAML, what do we need to produce?
> 
> I'm less interested in seeing a "widget toolkit" than I am in seeing  
> a killer set of library functions that can be drawn upon to build  
> custom behaviour. While some folks may want a stock widget set that  
> provides buttons and other form widgetry, tabsets etc I'm personally  
> more interested in being able to quickly build slick widgets "made  
> for Web pages", akin to Dave's (Jot) invite users widget for example.
> 
> I see this as having a solid set of discoverable and useful functions  
> (which also means documented) which are interoperable with other  
> libraries, lightweight and little hassle to develop with.
> 
>> What should we be incorporating from MochiKit and the other  
>> contributed code bases on the 0.2 and 0.3 timeframes?
> 
> I think it is much more valuable to interoperate with these other  
> toolkits than to try and suck them in.
> 
> What code are you thinking of when you ask this?
> 
>> And when should 0.2. and 0.3 be released?
> 
> I think we shouldn't aim too large for 0.2. I think there is more to  
> be gained from smaller and more frequent releases.
> 




More information about the NG-DHTML mailing list