"device independence"

From: David R. Karger <karger_at_theory.csail.mit.edu>
Date: Mon, 9 May 2005 13:16:49 -0400

I woke ope this weekend feeling like we've gotten very wrapped up in the "design"
of fresnel while I still feel unsure about the "requirements" of fresnel. So I
want to back up to that question, and think about what we actually want fresnel to
let us do, and even perhaps come up with some use cases.

I know stefano is sceptical of any "device independence" for fresnel. I'm still
thinking about whether it is possible. If we do want some amount of device
independence, it seems to me that we must offer one of two things:

1. Our "purpose" vocabulary must accomodate some notion of the capabilities of
   the device; eg, one must be able to ask for a lens able to display things
   in very small amount of space, or using only text, etc.

2. The lens that everyone uses must incoroporate some notion of relative
   importance of things, eg that the name matters a lot more than the address,
   so that a device/application can make informed choices based on its limitations;
   eg a cell phone might only show the most important property.

choice 1 has the advantage of being "not fuzzy". Everything is precisely defined.
On the other hand it calls for a lot of work on the part of the lens designer, and
probably requires a new lens for every device/application with different features.
In the end it gives less information, and thus less control, to the device.
Choice 2 forces us to think about fuzzy notions of greater or less importance.
But it should let us produce one lens that can be used effectively by a broader
variety of applications.

Right now I am leaning towards (2). What do others think?
Received on Mon May 09 2005 - 17:15:37 EDT

This archive was generated by hypermail 2.3.0 : Thu Aug 09 2012 - 16:40:51 EDT