I think that the performance of selected operations is
still a undefined area. Transcoding might
take place on an external (e.g., iCAP) server
and might take place in the proxy box.
In the proxy box I see this as an opportunity
for a library package - again keeping the
proxylets simple.
-----Original Message-----
From: Markus Hofmann [mailto:hofmann(_at_)bell-labs(_dot_)com]
Sent: Sunday, December 03, 2000 7:36 PM
To: Michael W. Condry; Nick(_dot_)Taptiklis(_at_)gregory(_dot_)co(_dot_)nz
Cc: ietf-openproxy(_at_)imc(_dot_)org
Subject: RE: Modified Example Services Draft
Michael,
This might be a good candidate for a special library.
Do you mean a library running on the proxy cache itself? Transcoding
is pretty processor consuming (especially transcoding streaming
media!) and you probably don't want to run it on the proxy itself. I'd
rather say that's a nice example for a remote-call out service.
The rule model
seems to require inputs from the primary data stream (like the web)
and the "user" and "content owner". This is a vector off the
"user" and primary data stream - allowing this to customize it
depending on the user's particular needs rather than a general
disability group.
Agreed - and it can be captured very nicely with our proposed PSRL
language :)
-Markus