Re: Implementing NodeList

>I completely agree with Steve on this. I'm using the DOM on a server
>side application with extremely large HTML document trees. Despite the
>current DOM spec's notice about the DOM not being usable in a multithreaded
>environment it really has to be on the server-side when many clients
>are accessing the same document at once. Otherwise resource usage becomes
>intolerable. To make NodeList and HTMLCollection always live is
>really asking for non-standard implementations. Maybe there needs to
>be a 'Scripting DOM' for the browser implementors and a spec for
>everything else that lends itself to somewhat efficient implementations.

All I can say is that your point of view is very well represented in the DOM
WG and IG.

>I'm fairly certain that the most intensive use of the DOM will
>not be for client-side special effects but on the server for
>more interesting document services via servlets et al.

I am a proponent of server-side DOM as well but it is my assessment that
client-side DOM is and will continue to be the most hyped use of the DOM
API.  Netscape and Microsoft has a chance to unify the their respective
browser's document model in Navigator 5.0 and IE 5.0.  It makes a lot of
sense not to lose sight of the potential benefits to the entire Net
community if this happens.

It will take a lot of time and frustration for the server-side DOM vendors
to pull together to scream together in harmony.  Just give it time.
Meanwhile, we will have to make do with implementation specific concurrency
control.  If you do not want to wait, gather some people and companies with
similar interest and start writing down the server-side DOM requirements.

Best wishes,

Don Park
CTO/Docuverse

>P.S. Am I correct in assuming that the usage of 'long' in the
>HTML DOM Java interfaces should really be 'int' ?


That is my understanding.

Received on Tuesday, 28 July 1998 02:34:36 UTC