cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefano Mazzocchi <>
Subject Re: [PROPOSAL] Cocoon Science Fiction
Date Tue, 11 Feb 2003 11:39:47 GMT
Niclas Hedhman wrote:

> However, I like the notion that there are "typing information" (of DTD/Schema) 
> for block's inputs and outputs, so at least the configuration tools (not the 
> sitemap in runtime, that's just waste of CPU resources) can validate the 
> "pluggability" between blocks.

I'll ask this 'at nauseam': how are you going to define those schemas 
between pipeline stages? At least you need a infoset-orthogonal 
validation language, otherwise, the fact that you validate or not will 
change your outcome. This forces you on the RelaxNG camp.

Now, suppose you have a stylesheet that transforms MVL (my vector 
language) into SVG and keeps everything else untouched. Then you have a 
generator that doesn't spit MVL at all. The two combine perfectly, yet 
it's silly to do so.... but how in hell are you going to find out?

I'm pretty sure some megaguru like Mr. Clark might be able to create an 
algebraic representation of the input and output schemas, than provide 
an permutation language to obtain 'matching' of the two dealing with 
multidimensionality of namespaces.

But I challenge anybody (megaguru included) to try.

> The exact mechanism for this is a lot harder, because it needs to be simple 
> and non-intrusive. Also, until there are blocks and some more solid 
> configuration tools, this is less important than, for instance, flows, and 
> can wait.

Oh, that's for sure.

Stefano Mazzocchi                               <>
    Pluralitas non est ponenda sine necessitate [William of Ockham]

To unsubscribe, e-mail:
For additional commands, email:

View raw message