[Xpipe-users] W3C's XML Pipeline Definition Language

[DuCharme, B. (LNG) <bob...@le...>]

I finally had a chance to read through the W3C XML Pipeline submission.
Because there was no location for public comment included, and because most
of the issues are related to XPipe, I thought I'd post my impressions here.
I'd like to hear what other people have to say.

Impressive list of submitters. 

1.4, the classification of processes into five classes was interesting. 

I like the processDef element, which "associates a name with an external
process definition." If a program that performs a task is upgraded or
replaced, only one pointer to it needs to be fixed, and the p:process
process definitions that use it can be left alone.

2.4.2.3: "It's an error for more than one process to produce the same
information set." How is that error caught? If process p1 turns infoset i1
into i2, and p2 turns i2 into p3, and p3 can turn i1 into i3, then processes
p2 and p3 are producing the same information set. Theoretically, p3 is
unnecessary, but practically, if p3 is being replaced by p1 + p2 or vice
versa, a period in which they're both in production is natural. 

My main issue with the Note is the processing dependency on infosets not
being up-to-date. I certainly understand the value of this make-like
checking of date-time stamps to determine which processes don't need to be
redone, but I'm not convinced that it deserves to be a part of the Pipeline
Definition Language, especially after reading the use cases. If this was not
in the spec (as it's not in XPipe) the pipeline requirements would be
simpler. Is there any reason a date-time stamp couldn't be passed in a
p:param element, giving the process the make-like time stamp to use in its
processing logic if desired while leaving the design of pipeline controller
simpler and cleaner (as it is in XPipe)?

Bob DuCharme          www.snee.com/bob           <bob@  
snee.com>  "The elements be kind to thee, and make thy
spirits all of comfort!" Anthony and Cleopatra, III ii