On Thu, Jan 30, 2014 at 10:03 AM, Dean Jackson <email@example.com> wrote:
We also have to be careful that this information doesn't become a UA-like string. i.e used for feature detection. That's why I think it should be specified - so that we all implement it the same way.
I'm not concerned about feature inference from an opaque string per-se. Unless you can detect the features working and the ones not, then that's the only thing you can do. I'm concerned about two things:
1) This being unspecified, so it's necessarily fuzzy, which is bad for developers.