Re: [syndication] Re: Thoughts, questions, and issues.

[Gabe Beged-Dov <begeddov@jfinity.com>]

Eric Bohlman wrote:
> 
> That bugged me too.  You shouldn't have to explicitly code simple
> parent-child relationships, any more than you should have to supply
> explicit counts for the number of items.  Aside from being unaesthetic,
> it's error-prone.  One of XML's biggest strengths is that it can easily
> model containment relationships.  <inchannel> looks like an explicit key
> you'd use if you were trying to model the data with an RDBMS.

The RSS1 proposal tries to serve multiple masters using a single XML
syntactic representation. It is very close to being syntactically
equivalent to RSS09 which was RDF based. One of the goals of RSS1 is
to allow both XML processors and RDF processors to grok the document
format with the emphasis being on XML processors first and RDF
processors second. The intent is also to maintain the balance between
simplicity and power. This is an always challenging task which as you
mention in another message is dependant on your context and what the
alternatives are.  

On the RDF front, RSS09 does not have any linkage between the channel
element and the item elements other than them being in the same file.
It also doesn't have item ordering information other than that of the
document ordering of the items in the file. It seems obvious that the
information is there in the file format, but as you note it is
analogous to the fact that rows in a relational database must be
explicitly tagged with ordering information. RDF is a semantic
representation and the XML/RDF encoding doesn't use document order to
infer semantic order. We have to explicitly encode that. 

You could argue this as being redundant, as you could argue alot of
different aspects of the XML encoding itself as being redundant.
Still, both XML+namespaces and RDF are W3C standards that we wanted to
adhere to in the RSS1 proposal while remaining as true as possible to
the spirit and intent of RSS.

The RSS1 core is simple enough to directly use as an authoring format.
It is also powerful enough to serve as an interchange format for both
controlled and uncontrolled extensions. An RSS_DW module can be
supported just as easily as an RSS091 and a consumer of RSS will
always be able to tell where the different bits and pieces come from. 

Some of us also believe in the semantic web and were willing to
concede a small amount of syntactic overhead in order to provide some
additional grist for the mill. If it turns out that overhead is simply
too much to bear we can consider falling back on the heart of the RSS1
proposal which is the use of namespace labelled modules to manage
extensions to the core RSS1 vocabulary.

Cordially from Corvallis, 

Gabe Beged-Dov

-- 
--------------------------- 
http://www.jfinity.com/gabe