- From: Martin Thomson <martin.thomson@gmail.com>
- Date: Tue, 10 Jun 2014 11:22:58 -0700
- To: Harald Alvestrand <harald@alvestrand.no>
- Cc: "public-media-capture@w3.org" <public-media-capture@w3.org>
On 10 June 2014 03:42, Harald Alvestrand <harald@alvestrand.no> wrote: > Do you have a different proposal in the works to figure out an exemplar > algorithm for how to resolve multiple "ideal" constraints together? I think that this is an area where we should permit a little innovation. I would recommend the following: 1. fulfill all ideal parameters that can be fulfilled 2. if there are none that can't be fulfilled, you are done 3. then either a. ignore any that remain, b. try to set the values to the "best" possible in an arbitrary order, or c. define an algorithm of your choosing (minimize sum of the squares of the differences to ideal, for example) to aid in selecting non-ideal parameters I think that 3.c. can be almost arbitrarily hard. I don't think that we need an exemplar. 3.b. might be good enough.
Received on Tuesday, 10 June 2014 18:23:28 UTC