On May 19, 2006, at 3:48 AM, P T Withington wrote:

> On 2006-05-19, at 00:40 EDT, Jim Grandy wrote:
>
>> One thing to be careful of is that making LzNode derive from  
>> LzDataElement would impose more restrictions on our ability to  
>> optimize LzDataElement by decorating the native XML data element  
>> types.
>
> Seems like there are two conflicting uses here:  1) making the node  
> hierarchy navigable and 2) binding the node hierarchy to data.
>
Exactly. But we don't need to over-use the class hierarchy for this.  
Keep in mind that we can have two (or more) separate classes/traits  
implementing the same "interface" in different ways. One for XML DOM  
trees, another for our node trees, for example. This is what I was  
getting at below with reference to Oliver's LzNode DOM apis.
>> Another way to get (almost) the same effect would be to adopt  
>> Oliver's DOM APIs for LzNode -- the ones in incubator/domapi.js.  
>> Not sure how this resonates with DHTML, though...
>
> You'd have to expand on this.
>
>> One way to resolve the cheasyMultipleInheritance problem would be  
>> to create a trait that implements the parts of LzNode that  
>> LzDataset wants, and have both LzNode and LzDataset derive from  
>> that trait.
>
> Henry and I are looking at making LzDataNode and LzDataElement be  
> traits and mixing them in, rather than trying to abstract a trait  
> from LzNode.  Do you think this is the wrong way to go?
>
No, I think that's a good direction. I wouldn't literally make them  
traits -- I'd refactor traits out of them, and have them derive from  
the new traits.
> One issue is that LzDataElement needs to be both a trait and a  
> class (doesn't it?), so we are probably looking at at least  
> abstracting out functionality from LzDataElement.  But that seemed  
> like a smaller task than trying to abstract LzNode just now.
>
Agreed.
>> On May 18, 2006, at 12:50 PM, P T Withington wrote:
>>
>>> This has been proposed several times.  I think one fear is that  
>>> it adds yet more weight to LzNode.  But both Jim and Adam have  
>>> discussed it in the past.  Perhaps now is the time to bite the  
>>> bullet.  Or, perhaps we make LzDataElement a trait?
>>>
>>> On 2006-05-18, at 14:03 EDT, Henry Minsky wrote:
>>>
>>>> Or rather, maybe LzNode should inherit from LzDataElement?
>>>>
>>>>
>>>>
>>>> On 5/18/06, Henry Minsky <[EMAIL PROTECTED]> wrote:
>>>>>
>>>>>
>>>>> What would be wrong with just having LzDataset inherit from  
>>>>> LzDataElement?
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Henry Minsky
>>>>> Software Architect
>>>>> [EMAIL PROTECTED]
>>>>>
>>>>>
>>>>
>>>>
>>>> -- 
>>>> Henry Minsky
>>>> Software Architect
>>>> [EMAIL PROTECTED]
>>>> _______________________________________________
>>>> Laszlo-dev mailing list
>>>> [email protected]
>>>> http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
>>>
>>
>

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

Reply via email to