I started wondering this morning whether
that might be scope for a "friend" class that
could extract this kind of stuff from nodes.

It would solve the bloat issue - you could keep
the source tree stuff lean and mean and have a
(set of?) helper classes that would do this work
for you by reading directly inside the nodes.

The (*huge*) downside is that you now risk having
the same issues that currently exist with the
wrapper/bridges.  You have a set of wrapping
classes that you need to support and keep up to
date.  The wrapping classes could potentially
get quite complex, particularly if you start
worrying about threading.  So I think I convinced 
myself out of it.

Cheers,
   Berin

> 
> From: David N Bertoni/Cambridge/IBM <[EMAIL PROTECTED]>
> Subject: RE: Should all of the unimplemented DOM functions throw exceptions?
> Date: 26/02/2003 7:55:47
> To: [EMAIL PROTECTED]
> 
> 
> 
> 
> 
> Hi Brian,
> 
> Yes, for small documents it's not a problem, but for large markup-heavy
> documents, it's adds up really fast.  If you can layer it, that would be
> great.  We could also make that something controlled by a #define, so if
> that would help you, it's OK with me.
> 
> Dave
> 
> 
> 
>                                                                                      
>                                                           
>                       Brian Quinlan                                                  
>                                                           
>                       <[EMAIL PROTECTED]         To:      [EMAIL PROTECTED]          
>                                                      
>                       com>                     cc:      (bcc: David N 
> Bertoni/Cambridge/IBM)                                                   
>                                                Subject: RE: Should all of the 
> unimplemented DOM functions throw exceptions?                    
>                       02/25/2003 12:39                                               
>                                                           
>                       PM                                                             
>                                                           
>                       Please respond                                                 
>                                                           
>                       to xalan-dev                                                   
>                                                           
>                                                                                      
>                                                           
> 
> 
> 
> > We already have a lazy implementation for that,
> 
> Ah, I see what XalanNodeListSurrogate does now.
> 
> > but I worry about bloating
> > the size of elements if we decide to add it, since it means 4 or 8
> extra
> > bytes per Element for an interface we don't use internally.
> 
> Added 10% to the size doesn't seem important to me but I don't work with
> huge source documents.
> 
> > Is it something you can layer on top of our implementation, or is that
> 
> > too awkward?
> 
> I can layer it on top of Xalan. It's a bit of a hassle but not a huge
> one.
> 
> Cheers,
> Brian
> 
> 
> 
> 

This message was sent through MyMail http://www.mymail.com.au


Reply via email to