Re: data API retransmission model

On 20 May 2014, at 16:51, Adam Bergkvist <adam.bergkvist@ericsson.com> wrote:

> On 2014-05-18 20:26, Michael Tuexen wrote:
>> On 18 May 2014, at 19:41, Cullen Jennings (fluffy) <fluffy@cisco.com> wrote:
>> 
>>> 
>>> If both maxPacketLifeTime and maxRetransmits are set, it should probably raise an exception as this does not seem t be allowed.
>> Correct.
>> 
>> It should also be specified that an exception (and which one) is raised if
>> * the label argument of createDataChannel() is too long.
The label can't be longer than 65535. The label length in
http://tools.ietf.org/html/draft-ietf-rtcweb-data-protocol-05#section-5.1
is only 2 byte.
>> * the protocol field of the RTCDataChannelInit is too long.
The label can't be longer than 65535. The protocol length in
http://tools.ietf.org/html/draft-ietf-rtcweb-data-protocol-05#section-5.1
is only 2 byte.
> 
>> * the id field of the RTCDataChannelInit is too large.
If it is the SCTP stream ID, it can't be larger than 65534.
See http://tools.ietf.org/html/draft-ietf-rtcweb-data-protocol-05#section-3

However, the number of streams could be negotiated to a smaller number by using
http://tools.ietf.org/html/draft-ietf-mmusic-sctp-sdp-06
However, http://tools.ietf.org/html/draft-ietf-rtcweb-data-channel-09#section-6.2
says that the maximum number SHOULD be negotiated. So there is a hard limit and
possibly lower ones depending of the association setup.
> 
> Michael, can you provide the valid ranges for the above (or a link to where that info can be found)?
See both above.
> 
>> * the id field of the RTCDataChannelInit is already used.
> 
> from spec: "Otherwise, if the value of the id attribute is taken by an existing RTCDataChannel, throw a ResourceInUse exception and abort these steps."
Great. Wasn't aware of this...

Best regards
Michael
> 
>> * the id field of the RTCDataChannelInit is incompatible with the odd/even rule
> 
> /Adam
> 
> 
>