- From: Silvia Pfeiffer <silviapfeiffer1@gmail.com>
- Date: Wed, 6 Jan 2016 07:24:16 +1100
- To: Harald Alvestrand <harald@alvestrand.no>
- Cc: public-media-capture@w3.org
- Message-ID: <CAHp8n2k+4MZViTotu-Fc=PrezSgKLYNL5eOwf+bmRgO9cVX-Rw@mail.gmail.com>
Hmm, the web platform is typically generous in ignoring things that are surplus. In this case though it's a coding error, because it's a required constraint that will always fail, so should be discoverable during coding time. I could live with either approach. Best Regards, Silvia. On 5 Jan 2016 11:04 PM, "Harald Alvestrand" <harald@alvestrand.no> wrote: > This is a riff on Adam's issue about known but unsupported constraints: > What's the expected behavior for constraints that are simply not > relevant to the case at hand? > > I'm thinking, for instance, of: > > getUserMedia({video: { echoCancellation: {exact: true}}}) > > echoCancellation is a known and supported constraint, but for video, > it's just not there - and the user is asking for a failure if it can't > be applied. > > Is the correct response to return a promise resolved to an > "overconstrainedError" with the "constraint" field set to > "echoCancellation"? > > This is what is implied by the current text for "fitness distance": > > "If the constraint is required ('min', 'max', or 'exact'), and the > settings dictionary's value for the constraint does not satisfy the > constraint, the fitness distance is positive infinity." > > I'm pretty confident that's what the spec says now, but I want to make > sure we're all agreed that this is what we intended. > > Harald > >
Received on Tuesday, 5 January 2016 20:24:46 UTC