Re: Issue 231: DataChannel: short vs. long data types

While I agree the value is overly large, the implication then is when eventing an incoming DataChannel (and parameters as obtained from the wire) then the presumption would be that values >  65,535 for maxPacketLifetime / maxRetransmit are therefor illegal and the incoming DataChannel must be actively rejected (i.e. for containing an illegal value). Or the value exposed to the app developer is capped at the max ushort range. Pick your poison...


On September 2, 2015 at 1:57:27 AM, Peter Thatcher ( wrote:

More than 65,536 is a *lot* of retransmits
More than 65s is a long time to retransmit.

I don't think there's much need to make those 32-bit.  Why would anyone ever use values that big?  Plus, we might have a different protocol in the future which doesn't use 32-bit for them.  No point in being overly tied to SCTP.

On Mon, Aug 31, 2015 at 1:55 PM, Bernard Aboba <> wrote:
>From Robin Raymond:

dictionary RTCDataChannelParameters {
DOMString label = "";
boolean ordered = true;
unsigned short maxPacketLifetime;
unsigned short maxRetransmits;
DOMString protocol = "";
boolean negotiated = false;
unsigned short id;
unsigned short maxPacketLifetime;
unsigned short maxRetransmits;
^ those aren’t shorts (16 bits), they are 32 bit in the protocol

Received on Monday, 14 September 2015 17:04:02 UTC