Re: What would you like to see in WebRTC next? A low-level API?

On 1/24/2018 12:59 PM, Sergio Garcia Murillo wrote:
> On 24/01/2018 18:50, Michael Tuexen wrote:
>>
>> Are you referring to a non-congestion controlled way of sending 
>> unordered, unreliable datagrams?
>> You can right now use and unordered delivery, limit the number of 
>> retransmissions to 0 and
>> get a congestion controlled unordered unreliable datagram service...
>
> Except that it can be buffered, chunked, there is no way of knowing 
> the MTU and last time I checked didn't work as expected on any 
> browser.. ;) 

There is a transmit buffer, and it does hide the MTU from you, true - it 
provides an abstracted UDP API, not a raw socket API. If you could ask 
the current apparent MTU as seen by the protocol, you could restrict 
your messages to that.   But of course that same issue exists for UDP - 
you can send larger messages (ok, max 64K) and it will fragment and 
reassemble (maybe) the messages for delivery, just like DataChannels.

So that leaves exposing the (current) MTU, and the fact it can be 
buffered (though normal network IO can be buffered as well).  Would you 
be happy with exposing MTU?

-- 
Randell Jesup -- rjesup a t mozilla d o t com
Please please please don't email randell-ietf@jesup.org!  Way too much spam