Re: [rtcweb] Working group last call for draft-ietf-rtcweb-audio

[Roman Shpount <[e-mail]>]

On Fri, Dec 18, 2015 at 4:24 PM, Adam Roach <[e-mail]> wrote:

> On 12/18/15 15:10, Roman Shpount wrote:
>
>> It looks like these tones are removed for aesthetic reasons only.
>>
>
> I think it would be better if you directly disagreed with Cullen's
> rationale rather than simply ignoring it. Since it was made in a different
> forum, I'll copy it here.
>

I was reminded multiple times that here is the right forum to discuss this
issue. I am mostly looking for a consensus made here and then propagated to
W3C.


> On 12/9/15 12:48, Cullen Jennings wrote:
>
>> I want to be clear I don't care that much one way or the other about this
>> but I want to try again to explain the issue...
>>
>> A to D will often be blocked by the system that sends them from one side
>> to the other. So if you write JS applications that use theses, they will
>> work some times and not others. That is just inviting developers to shoot
>> themselves in the foot and no one has identified any benefit of adding A to
>> D yet.
>>
>> 100% agree that the IETF and W3C specs need to align on this and we
>> should change the ietf audio draft if we are including A-D.
>>
>>
> To be clear, this is speaking to operational aspects of the system's
> behavior that I haven't had much visibility into. We have one implementor
> effectively asserting that all commercial gateways support these tones, and
> another saying that they tend to be blocked by intermediaries. I suppose
> both assertions could be true, but the net effect of both being true is
> that these tones are broadly unusable.
>
> But I'm in the same boat as Cullen: I don't care much whether these tones
> are supported, as long as the W3C and IETF documents agree with each other.
>

DTMF tones A-D can be tricky and might not work in 100% of cases.
Unfortunately this applies to all DTMF tones. For instance, DTMF tones from
the party receiving the call do not work in a fair number of cases. If
developer is using DTMF tones, they should be prepared to deal with
consequences. For instance, as a part of operating a conferencing service
with traditional PSTN access we need to deal with 2-3 tickets per day which
are related to DTMF transmission over various telephone networks.

At the same time WebRTC application can be connected to a legacy PSTN
system under developer control (i.e. through the gateway to a PBX which
uses tone A-D for special purpose such as to trigger phone line flash). In
such application these tones will work every time with 100% certainty.

In my previous email I have listed examples of legacy applications that
still use DTMF tones A-D. I think this is reason enough to keep them.

My main point is that it is probably easier to leave the legacy things
alone instead of improving them. DTMF should not be used for any new
applications. Datachannels are a much better option. DTMF should be left
for legacy interop only, so there is no reason to redesign them or make
them easier to use. They worked fine for the past 50 years.
_____________
Roman Shpount