xml-dev - Re: Procedural vs Declarative XML transformation approaches

[Paul Tchistopolskii <paul@qub.com>]

From: Thomas B. Passin <tpassin@home.com>

> Are you and Matt really thinking about mixed content (text
> and elements) or are you thinking about ***unknown***
> combinations - that is, there is a lot of flexibility about
> what can come next?

I agree with you also. ( too bad you've captured this flaw, 
because it requires me to say something I hate to say. )

You are right. The *real* problem for pull processing is 
'messy content' ( the entity which is missing in any 
XML document I've seen. I wish I'm wrong and 
there is some W3C paper which talks about 
'messy content' ? I appreciate the URL ).

Mixed content is only a part of bigger usecase - 
'messy content'. This was concluded on SML-dev list 
like ... one year ago...

Unfortunately, making the next step requires
discussing other issues that are not related to 
declarative/procedural in XSLT.

We are better not to open this can. 
( Just to explain - for example, *some* cases of 
'messy content'  are better to be processed with 
saxon:evaluate == mutable stylesheets. ).

Because this has nothing to do with Matt's point about 
declarative / procedural part of XSLT,  I thought that 
I'm better to stay in the borders defined by W3C.

I mean :

'Mixed content' is legitimate entity.

Matt's sentence regarding 'mixed content' is correct.

In the real life people rarely use messy content 
without mixed content ( it is possible, but .. ) - 
so let's simplify things a bit and agree with Matt, 
because what he is saying is very reasonable.

Discussing 'messy content' is not safe thing.

Thanks for understanding.

Rgds.Paul.

PS. BTW - my other point remains valid. The problem 
domain of XSLT is to render documents : 'messy content'.
Mostly 'mixed content'. 

Other usecases may require another balancing of 
declarative / procedural.