Re: A "processing model" proposal

> So is the following generalization sufficient?
> 
> Where I had one anonymous infoset, you propose a list of anonymous
> infosets. Each step consumes all of them (perhaps by ignoring some of
> them) and produces zero or more.

Provided that you allow for the multiple output infosets from one step
may be consumed by different steps.  I also want a single output to be
connectable to multiple steps' inputs, but we could have (conceptually
at least) a "tee" component that produces multiple copies of its
input.

> You also have "local names" which I'm on the fence about, but I think
> that's a seperable issue.

The local names are merely a mechanism to allow the graph to be
represented in XML; they are irrelevant to the processing model
itself.  I think we should explicitly separate the processing model
from the XML representation of the pipeline: the XML document
represents a graph of steps, and the processing model describes how
the graph is executed.

-- Richard

Received on Thursday, 16 February 2006 19:21:49 UTC