Re: [Gzz] PEG content_handler

[Tuomas Lukka]

On Fri, Apr 04, 2003 at 02:31:52PM +0300, Matti Katila wrote:
> ==========================================================================
> PEG content_handler--mudyc: Constructing simple handlers for specific span
> ==========================================================================
> 
> :Author:   Matti Katila
> :Stakeholders: Matti Katila
> :Last-Modified: $Date: 2003/04/04 11:18:30 $
> :Revision: $Revision: 1.1 $
> :Status: Incomplete  
> 
> 
> Abstract
> --------

This is just about the right size for a PEG. You're learning fast ;)

Comments included.

> This PEG is close to PEG nodeview_abstract--mudyc where I did
> suggest simple method to place a node without any information
> about the content of a node. 

This sentence doesn't help the reader - that it's close doesn't
tell me anything.

> Now, in this PEG, I've a new solution
> for the problem that we still don't know a way to handle placing
> part of the job.

This is not very clear: you are solving something differently 
than in the other PEG? Is some part of it outdated?

If not, why not start with what you're actually doing, instead of
putting it inside so many convolutions:

        This PEG defines a generic API to place an Enfilade1D
        into a coordinate system.


> Design
> ------
> 
> NodeView, which is responsible to place a node, must get information
> of the content of the node. Space/FFC provides this information with 
> method getEnfilade. Once we have enfilade we can split it smaller ones
> so that TextSpan, ImageSpan and PageSpan etc. are in their own enfilades.
> Now we can give the pieces of content to correct handler(text,image and page 
> etc.)
> where it is placed in a vob and putted in the vobscene.

Umm, this is a bit complicated also; the point is rather that we need a
proper interface for placing enfilades onto the screen. This may then be
implemented by special classes for different media types.


> Changes
> -------
> 
> We need a new interface for different content handlers:
> 
> ::
> 
>     public interface ContentHandler {
>         void place(VobScene vs, int box, Enfilade enf);
>         void getSize(Enfilade enf, float[] size);        
>     }

Issue: can this be efficient enough? Often, views will first ask for the size, 
and then decide
on into how large a box to place the stuff. 
Will caching work properly when enfilades are used as keys?

[now you should put this into the PEG into the Issues section and once you know 
a resolution,
repost the PEG]

Issue: how does this interact with Benja's structured graphics stuff? I think 
that might
be well worth considering here.

Issue: would it be appropriate to PEG something in alph about media types? E.g. 
separating
an enfilade into parts of different media types is pretty generic.

        Tuomas