Re: manifest based processing

On 19 February 2014 14:07, Romain Deltour wrote:

> b.  what I suggested (on tweeter) was not an extension of the
> "kind" attribute

  Interesting.  Actually I missed that part of Jostein's email, I have
just seen it now that I saw this mention in your own email.  What is
interesting is that I was writing a response to Jostein suggesting
exactly the same thing as he did: being able to define the "kind" of a
port, and being able to connect implicitly outputs ports to the inputs
ports of the same kind on the net step.

  In addition to that "use a kind in order to create it", which looks
like @mode in XSLT 1.0 and 2.0 (a mode exists as soon as you create a
template rule with a @mode with that name), there could even be a
p:kind declaration (like XSLT 3.0 introduces xsl:mode).

  This has 2 advantages: detecting typos (if you make a typo, a new
kind is not created but the compilation gives you an error), and
allowing to set properties on the kind itself (is there an opportunity
to set a content-type?, validate outputs and inputs based on a
schema?, asking the processor to log all documents flowing through
ports of a given kind, etc.)

  That would probably solve verbose explicit bindings between steps of
the same library or application, acting on the same (sets of) kinds of
documents.

  Regards,

-- 
Florent Georges
http://fgeorges.org/
http://h2oconsulting.be/

Received on Wednesday, 19 February 2014 13:27:20 UTC