RE: [Bug 22209] "each key MUST be a valid registered constraint name in the IANA-hosted RTCWeb Media Constraints registry"

OK, let’s go back to a statement of your from a previous email:


Ø  Why differentiate unknown from unsupported?

The constraints listed in Section 12 are known.  If an implementation supports a subset of them, they are known and unsupported.  I was trying to respond by saying that I didn’t think this was possible – a valid implementation must support all the known constraints outlined in the specification.

> I don't think you got my point

On retrospect, I agree.



From: Jan-Ivar Bruaroey [mailto:jib@mozilla.com]
Sent: Wednesday, November 06, 2013 3:12 PM
To: Mandyam, Giridhar; public-media-capture@w3.org
Subject: Re: [Bug 22209] "each key MUST be a valid registered constraint name in the IANA-hosted RTCWeb Media Constraints registry"

On 11/6/13 1:58 PM, Mandyam, Giridhar wrote:

This bug isn't about "required to implement" is it?



Thanks for responding.  I didn't write the original statement in the spec or the bug, so I can't answer.  But I do think there needs to be clarity as to what is the minimum set of constraints (if any) that a valid implementation should support.

I think there needs to be clarity in this conversation. Two issues are conflated here:

  1.  Bug 22209: "constraints [MUST|SHOULD] be IANA registered" - My answer: SHOULD
  2.  Your question: What's the min set of constraints required to implement? - My answer: No strong opinion.



I assumed the constraints listed in http://dev.w3.org/2011/webrtc/editor/getusermedia.html#iana-registrations constitute a minimum set of constraints that valid implementations MUST implement.  The group trimmed a list of nearly 30 proposed constraints to what is currently in the spec  based on broad consensus (peerIdentity and noaccess being exceptions).  If a valid implementation does not have to support even the constraints listed in the spec, then I wonder why the group went through the exercise of prioritizing constraints for the first version of the spec.



Saying we desire everyone to register their keys, is different from saying the API may crash or detect such a thing. MUST is a FUD-based empty mandate, while SHOULD conveys the nature of the mechanism more accurately IMHO. I care about clarity, and your confusion about implementation fuels my concern that the word MUST is the culprit (see below).



Got your point.  If this is the intention of the group, then I wonder why we bothered with the exercise of trimming constraints in the first place.

I don't think you got my point, because my answers were about Bug 22209, not your issue.

I don't have a strong opinion on your issue. Section 12 calls the constraints "stable", and we are working on implementing them. I remember the 22 others as mostly obscure and from a single source, so I found the triage extremely helpful to let us prioritize the important ones, regardless of what the spec ends up saying about them.

.: Jan-Ivar :.