I was thinking XSLT does a lot of pipe-like functionality.
At the meeting where this was brought up, someone asked about wether
Yahoo Pipes offered the functionality of being fed in a list of feeds,
e.g. an OPML file?
So I made a trivial template that transforms a list of URLs into table
on a web page, with the first few entries on each feed on each row.
http://xslt.portlanddatasystems.com/proc.php?_xml=feeds.xml
It can also be run on the client, in browser (with no server-side
processing other than the proxy to get around the Firefox same-domain
restriction)
http://xslt.portlanddatasystems.com/client_feeds.xml
That's just something pulled together from all the feeds each time the
page is requested, so it would't exactly be scalable for loading up
your opml of 900 feeds -- generally I was thinking you'd want to cache
normalized feeds into a database something through a REST interface
prefferably. I had done this with the typefaces OPML -- fleshed out
the outline, and put it into an XML database, but I don't have a web-
accessible Java runtime at the moment.
As a more direct fit and general-purpose app, there's a rather new
open standard XML Pipeline langauge "XProc" with a couple
implementations out.
XProc has an advantage of mapping directly to the pipes structure
(connecting boxes together with tubes). Also, existing pipelines can
be boxed up and used as boxes within other pipelines (wired-together
boxes as a box in another pipeline).
It does seem more general in scope than pipes, besides aggregating,
sorting, mashups, it can also be used as a general build sysytem,
interacting with databases, generating PDFs...
Common to both languages, is that they're written in XML, which would
make it simple to programaticlly parse / generate to and from the
visual layout. And they both can load additional modules from URLs,
so one could use portions of a friend's pipe anywhere on the net.
I'd like to get together with some folks and work on making pipe-
things!