Just to clarify:

Yes, namespaces, packages, and inner classes all sound like great  
extensions.  As to the DOM API's.

BUT, this proposal is only aimed at solving the immediate problem of  
collisions with browser globals.  Nothing in this proposal precludes  
all those other ideas, but I need to implement a short term solution  
now.

Regarding the DOM API's, I have a specific question for Max.  To add  
a child in the DOM API, you first createElement, then you  
setAttribute's on the element, finally you appendChild the element to  
its parent.  The question is:  can the Sprite API support this  
paradigm?  Or does the Sprite API depend on the child element being  
created and added to the parent in one step?  LzDataElement comes  
close to the DOM API, but is extended by us to let you specify  
attributes when you create the element.  I assume we would want to  
keep that extension.  LZNode goes one step further and lets you  
specify (a template for) child nodes at creation time too.  Do we  
want that extension?

If we can implement the DOM API's, then it probably makes sense to  
implement the NS (namespace) variants on these.  We would need  
additional syntax in LZX to take advantage of that.  Someone should  
propose (separately) what that would look like.

Since the <class> tag only defines additional tags, I see no reason  
to provide any special support for `new <class-that-implements- 
tag>`.  That is an implementation detail of createElement that should  
not be exposed.  [There is an outstanding bug for support of the  
Javascript 2 `class` declaration in <script> tags, once that is  
working, if you want an LZX-level class, as opposed to a new DOM  
element type, you would use that.]

If I really wanted to be incompatible, I would say that we should  
change the name of <class> to <element>, since it really is a  
specification of a DOM element type, not a class in Javascript.

I'm not ready to think about inner <class>'s.

_______________________________________________
Laszlo-dev mailing list
[email protected]
http://www.openlaszlo.org/mailman/listinfo/laszlo-dev

Reply via email to