W3C home > Mailing lists > Public > public-automotive@w3.org > September 2017

RE: wwwivi (was Re: WG call 2017-09-05)

From: Hyojin Song <hyojin22.song@lge.com>
Date: Tue, 12 Sep 2017 22:43:50 +0900
To: <ted@w3.org>, "'Gavigan, Kevin'" <kgavigan@jaguarlandrover.com>
Cc: "'public-automotive'" <public-automotive@w3.org>
Message-ID: <013501d32bcd$2aa5df00$7ff19d00$@lge.com>
Hi Ted,

I've before mentioned the RFC6762 in the github issue #223.
https://github.com/w3c/automotive/issues/223

I tried to read the spec carefully so that I think '.local' couldn't be used to replace with wwwivi. '.local' can be resolved to (local) IP address(IPv4, IPv6) while the mDNS ethernet frame from '.local' is a multicast UDP. I first think that '.local' can be resolved directly to the local address of WebSocket server, but it can't be. The reason is that it will be resolved to the mDNS response (UDP) packet in which the IP addresses(IPv4, IPv6) are specified. You can see the relevant example (appletv.local) as the following link.
https://en.wikipedia.org/wiki/Multicast_DNS#Example

In VISS spec, the following commend is specified to connect to a WebSocket server and get a vehicle instance.
var vehicle  = new WebSocket("wss://wwwivi", "wvss1.0");

If we'd like to replace 'wwwivi' with 'ivi.local', we can't use the commend above. Alternatively, we could define or find other method instead of the single WebSocket method, but it needs some time to do.

As I remember, we do not prefer to consider the discovery phase due to complicated implementation and overhead on the protocol for a thing, but I think it could be managed by using a simplified unified API for IVI so that the single method with 'ivi.local' makes a client connect to the WebSocket at once. It would be one of the candidates to solve this issue, and I'm okay to maintain 'wwwivi' for now as well, while we could consider new API as a future work.

Thanks,
Hyojin

-----Original Message-----
From: Ted Guild [mailto:ted@w3.org] 
Sent: Tuesday, September 12, 2017 6:09 AM
To: Gavigan, Kevin
Cc: public-automotive
Subject: Re: wwwivi (was Re: WG call 2017-09-05)

mDNS was an early response from my colleague in the TAG.

https://tools.ietf.org/html/rfc6762

We discussed, but not finding it in the minutes, discovery/broadcast in the past. Does anyone recall arguments against?

On Mon, 2017-09-11 at 16:03 -0400, Ted Guild wrote:
> Kevin had asked me to try to get a final resolution on this recurring 
> topic that has always been awkward for us. Hyojin has also suggested 
> escalating a decision.
> 
> As I recall Junichi got .localdomain (not reserved TLD) or .localhost 
> suggestion from Web of Things folks and that we would need to go with 
> self signed certs.
> 
> I sought advice from some colleagues who suggested I raise it to W3C 
> TAG which I have done. I will keep the group appraised and if they 
> want us to join a call with them will see who is interested in 
> attending from our side.
> 
> On Wed, 2017-09-06 at 15:05 +0100, Gavigan, Kevin wrote:
> > 
> > Re: https://github.com/w3c/automotive/issues/223
> 
> 
--
Ted Guild <ted@w3.org>
W3C Systems Team
http://www.w3.org
Received on Tuesday, 12 September 2017 13:44:19 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 24 October 2017 18:52:54 UTC