Implementing "Ideal" - implementation considerations (#118)

During the development of "constraints", the worry was raised that the
algorithm specified ("generate all combinations of possible
configurations and rank them according to fitness distance") couldn't be
implemented in practice.

https://github.com/w3c/mediacapture-main/issues/118

I wrote up a paper showing one way to do it given knowledge of the
internals of drivers and so on that allowed for reasonably fast
computation even when some of the variables were continuous rather than
discrete:

https://docs.google.com/document/d/1-rM0-kaBn2vOMm5Om8aZW6IF9pFIpm5385D48HUWzWA/edit

The question before the group is what to do with the spec given that
discussion.
The options I see are:

- Don't mention this in the spec, keep the paper around to show to
people if they complain about it
- Add the paper (suitably edited) as an appendix to the spec, so that
people who need it can easily find it
- Point out grave errors in the approach, leading to unsolvable issues
with "ideal", and let's start discussing what changes that implies

I'd like to have some feedback from the group on what course they see
reasonable.
(Comments on the note are welcome too!)

Harald

Received on Friday, 12 June 2015 08:38:41 UTC