[Fwd: how device independent?]

From: Stefano Mazzocchi <stefanom_at_mit.edu>
Date: Mon, 21 Mar 2005 12:56:33 -0500

Forwarded, for the same reason.

Stefano Mazzocchi
Research Scientist                 Digital Libraries Research Group
Massachusetts Institute of Technology            location: E25-131C
77 Massachusetts Ave                   telephone: +1 (617) 253-1096
Cambridge, MA  02139-4307              email: stefanom at mit . edu

attached mail follows:

[this message is long and one main point is at the end, so I'll lead
with it here: perhaps the only job of fresnel should be to specify how
to construct a new rdf model, containing the subset of the existing
model that we want to display (eg which properties) and everything
else should be left to the specific rendering device/application.]

We've been talking some about device/app-indepndent renderings. Our
data model is nicely independent. And we've been trying, in fresnel,
to produce a somewhat device independent description of how to present
information---eg, fresnel:showProperties is very much about display,
but not particular to a device. But now it seems we have two options:

1. we can give this howto language to each device, and allow that
   device to use it to create a good rendering. in this case the
   device would probably look directly at the data model (as well as
   the lens instructions) in producing the rendering

2. we can try to use the lenses, at display time, to create a
   relatively device independent description of what to present, that
   nonetheless separates the device (or app) from a need to query the
   data model directly. i.e., we could use the lenses to unpack the
   data model into a collection of literals (including strings, data,
   images) that need to get displayed, and then pass those literals,
   properly tagged with info like <fresnel:value> that can be used by
   the device to figure out how to display.

Option 2 is clearly much harder, and might not even be desirable. I'm
not sure at this point whether we are advocating it or not. My sense
is that we are trying to do it a bit, in that we are producing XML but
pushing out certain (presumably device dependent) stylings into the
css. There is some amount of attempt at device independence in the
xml representation of chris' last email, but (suggesting that actually
we are aiming for 1.) at the same time there is explicit reference to
certain rdf predicates in the data model.

If we plan to allow the renderng device to be smart enough to figure
things out about those predicates, then perhaps we should just go all
the way and serialize all the rdf we are displaying into rdf-xml that
we deliver to the rendering device. Not being facetious here: one
could argue that the only job of the device independent part of the
system is to figure out a SUBSET of the rdf model, which is itself an
rdf model, and deliver it to the rendering device, whose only job is
to create a rendering of ALL the rdf in the model it receives. This
is perhaps asking a lot of any rendering device, and abandoning a lot
of opportunity for solving certain aspects centrally, but it does
offer a very clean separation of concerns, and there is still a role
for fresnel and lenses to tell us how to construct the rdf subset
model for delivery to the device.
Received on Mon Mar 21 2005 - 17:55:00 EST

This archive was generated by hypermail 2.3.0 : Thu Aug 09 2012 - 16:39:17 EDT