Re: [rtcweb] RTCWEB Data Channel: Usage of PPID for protocol multiplexing

[Paul Kyzivat <[e-mail]>]

On 2/6/14 6:19 PM, Harald Alvestrand wrote:
> On 02/06/2014 09:47 PM, Michael Tuexen wrote:
>> On Feb 6, 2014, at 9:35 PM, Christer Holmberg <[e-mail]> wrote:
>>
>>> Hi,
>>>
>>>>>>> I hope that if values are retained simply for backward compatibility they are explicitly marked as deprecated.
>>>>>> My plan was just not to defined what they are used for...
>>>>> I would like to have some explanation text. Having registered values without any kind of explanation is not a good approach, in my opinion.
>>>> The reason why I registered them is that they are used by Firefox and there was
>>>> the possibility that the PPID based fragmentation and reassembly got
>>>> standardised. I wanted to avoid the case that Firefox uses some values
>>>> and other protocols register these values. That would have been pretty
>>>> bad. Since they are a very cheap resource, I just went ahead and "saved"
>>>> the values.
>>> That is a all good. But, when people wonder 5 years from now, I don't want them to have to look for this e-mail in oder to get an explanation :)
>> Understood. But do you think they walk through the IANA registry? I would expect them
>> to read RFCs and follow the procedures described there.
>
> Updating the registry to read "Obsolete; was used by <draft name>"
> should satisfy all comers.
>
> People do read registries.

Yes. In some cases the registry is how you discover what RFC to read.

Maybe five years from now, somebody debugging some app gets one of these 
values, and doesn't know what it means. It won't be mentioned in any of 
the current RFCs he has read, because none of them use it.

So he ends up looking in the registry to find where it is defined.

	Thanks,
	Paul