Having used SSP extensively in Swiki.net and having looked at Seaside, I
think I can offer some insight:

My sense is that both approaches are equivalent if you raise your level
of abstraction such that you are thinking in terms of "widgets" and not
HTML or an HTML DOM.

For example, I have a widget called RolloverButton (which has a mouse
rollover effect).  That rollover button can just as easily general html
indirectly using a DOM approach, or more directly with an SSP approach.
The important thing is that you have a RolloverButton and use it (and
other widgets) instead of directly dealing with HTML or an HTML DOM to
construct your UIs.

If I could do it all over again, I would probably go the DOM route.  If
I constuct the HTML indirectly through a DOM, then I could conceivably
generate other languages (not just HTML) out of that DOM.  On the other
hand, it might be almost as easy to have the widget directly render the
different languages. 

- Stephen 

> -----Original Message-----
> From: squeak-dev-admin at lists.squeakfoundation.org 
> [mailto:squeak-dev-admin at lists.squeakfoundation.org] On 
> Behalf Of Eric Scharff
> Sent: Wednesday, December 11, 2002 7:29 PM
> To: squeak-dev at lists.squeakfoundation.org
> Subject: Re: Template mechanisms...
> 
> 
> I happen to prefer the "pure" template mechanism of 
> WebObjects (and presumably Enhydra's XMLC) myself.  This is 
> the style I've tried to use given the Swiki's template mechanism.
> 
> One of my biggest problems with (my understanding of) the 
> "code-based" approach taken in SeaSide is simply that I end 
> up with large literal strings embedded in code, a style I 
> despise even more than the opposite (the JSP style).  My 
> spelling is terrible, so I like templates that I can 
> spell-check, and spell-checking the Smalltalk code directly 
> is usually problematic.
> 
> Is there a decent technique for avoiding having huge string 
> literals in the "direct" code approach?
> 
> -Eric
> 
> 
> 
> 
>