IETF
core@jabber.ietf.org
Tuesday, July 21, 2015< ^ >
irob has set the subject to:
Room Configuration
Room Occupants

GMT+0
[11:31:27] jaimejim joins the room
[11:54:40] jaimejim leaves the room
[13:15:27] kaorumaeda joins the room
[13:17:02] metricamerica joins the room
[13:17:18] Meetecho joins the room
[13:17:51] Barry Leiba joins the room
[13:19:36] metricamerica leaves the room
[13:21:14] jaimejim joins the room
[13:22:31] Robert Cragie joins the room
[13:23:36] <Robert Cragie> If you have a question for the microphone, please preface with "mic:"
[13:24:03] Benjamin Black joins the room
[13:24:27] HS YANG joins the room
[13:25:07] Dejene Oljira joins the room
[13:28:21] Robert Cragie leaves the room
[13:28:28] Robert Cragie joins the room
[13:30:05] Lucas Jenss joins the room
[13:30:24] Thomas Watteyne joins the room
[13:30:37] Dejene Oljira leaves the room
[13:37:44] <Benjamin Black> if you're asking, i read the draft.
[13:38:36] <Benjamin Black> i vote for #1 since it is the simplest.
[13:40:53] <Benjamin Black> the reliability was not put in udp, it was put in the coap header.
[13:41:55] <Benjamin Black> seems nobody is reading jabber.
[13:42:10] <jaimejim> I am ;)
[13:43:00] <Benjamin Black> the illusion of participation.
[13:43:50] <Benjamin Black> definitely need to define a fixed header format instead of redefining it in every transport.
[13:43:52] Pascal Thubert joins the room
[13:44:13] <jaimejim> that is a good comment
[13:44:19] <jaimejim> should be asked in the mic
[13:44:28] <Benjamin Black> how do you handle reliable delivery through a tcp proxy to udp?
[13:46:24] Pascal Thubert leaves the room
[13:46:41] <Robert Cragie> If you specifically want a question to the mic, please say so
[13:46:45] pascal thubert joins the room
[13:46:45] <Robert Cragie> I will relay
[13:46:52] <jaimejim> Remember to preface with “mic:”
[13:46:52] <Lucas Jenss> I thought a Jabber scribe had been selected, but they do not seem to be paying attention. If you'd like something to be brought to the mic please prefix it with [mic] and I'll try to get it there.
[13:46:53] <Benjamin Black> see above.
[13:46:55] <Lucas Jenss> oh
[13:46:57] <Lucas Jenss> too late :D
[13:47:00] <Robert Cragie> I am paying attention
[13:47:19] <Lucas Jenss> My bad then. No offense, sorry.
[13:47:23] <Benjamin Black> mic:definitely need to define a fixed header format instead of redefining it in every transport.
[13:47:44] <jaimejim> Anyhow, we could also discuss in the jabber in the meanwhile.
[13:47:48] <Benjamin Black> [mic]how do you handle reliable delivery through a tcp proxy to udp?
[13:48:58] <Robert Cragie> TCP would trigger a CON in the proxy to extend the reliablity (paraphrased)
[13:49:51] <Robert Cragie> Andrew MacGregor suggests still using CON/ACK
[13:49:58] 12312332 ddd joins the room
[13:50:07] <Benjamin Black> [mic]this is yet another side effect of not having a clearly defined coap header independent of transport. can we please define a fixed message header and use it for all transports?
[13:51:32] <Benjamin Black> [mic]why proliferate the problem with every transport?
[13:51:46] <Robert Cragie> Carsten Bormann says no, as that would mean having to change the existing header
[13:52:09] <Benjamin Black> we can leave udp alone because it is already defined
[13:52:25] <Benjamin Black> but tcp and websockets define different headers
[13:52:33] <Robert Cragie> I don't think there is any intention to change the header for UDP
[13:52:39] <Benjamin Black> yes i know
[13:52:51] <Benjamin Black> i just said leave it alone, but for all new transports use a single header
[13:53:06] <Benjamin Black> tcp and websockets define _different_ headers
[13:53:11] <Robert Cragie> So are you suggesting a single alternative header for all other transports?
[13:53:16] <Benjamin Black> yes
[13:53:23] <Robert Cragie> OK, got it
[13:53:38] <jaimejim> How do you know that that new header fits all other possible transports?
[13:53:46] <jaimejim> (to raztu)
[13:53:58] <Benjamin Black> because the new header is ONLY coap
[13:54:27] <Benjamin Black> there are transport fields mixed into the app header in udp coap
[13:54:40] <Benjamin Black> split them apart and then have a per transport shim
[13:54:55] <Robert Cragie> I think that is a good idea
[13:54:56] <jaimejim> I see. Worth asking to the mic IMHO
[13:55:05] <Robert Cragie> Shall I put it to the mic?
[13:55:10] <Benjamin Black> please
[13:55:11] <jaimejim> yes please
[13:55:27] 12312332 ddd leaves the room
[13:55:28] <Benjamin Black> i've asked on the list and tim wrote a whole draft on it, but carsten refuses to budge
[13:56:35] <Benjamin Black> agree it should be a wg item
[13:56:47] <Robert Cragie> Which draft is that (for the record)?
[13:56:50] Rosen, Brian joins the room
[13:57:26] <Benjamin Black> the documents should not be merged
[13:57:42] <jaimejim> call for a hmmm
[13:57:47] <jaimejim> humm
[13:58:45] <Robert Cragie> OK, I guess we are discussing this now
[13:58:50] <Benjamin Black> yes, good
[13:59:04] <jaimejim> yep
[14:02:49] test test joins the room
[14:03:56] test test leaves the room
[14:09:52] <Benjamin Black> [mic]this is not about re-using NON and CONs, it is about having clearly defined primitives.
[14:10:08] <Benjamin Black> the suggestions do not require changing coap.
[14:12:39] renzoe joins the room
[14:14:28] Mike Jenkins joins the room
[14:14:39] jaimejim leaves the room
[14:14:44] <Benjamin Black> this is an amazing discussion. is someone _not_ at a university arguing in favor of keeping the primitives undefined?
[14:15:56] <Benjamin Black> [mic] could someone objecting to defining primitives as tim describes address the substance of his argument? the slide that shows every transport having to define how observe and pubsub and future additions will be supported per transport?
[14:17:00] Jaime Jimenez joins the room
[14:18:22] Borje Ohlman joins the room
[14:19:02] Jaime Jimenez leaves the room
[14:20:33] <Robert Cragie> Is Klaus answering your question?
[14:21:38] mmani joins the room
[14:21:54] Paventhan Arumugam joins the room
[14:22:43] akeranen joins the room
[14:23:50] Mike Jenkins leaves the room
[14:26:31] M Jenkins joins the room
[14:32:16] <Benjamin Black> [mic] the primitive is the same in both those cases.
[14:32:35] <Benjamin Black> [mic]this is exactly tim's point. define the primitives and how each transport provides them.
[14:32:53] <Robert Cragie> Is this on slide 50?
[14:32:57] <Benjamin Black> yes
[14:33:30] <Benjamin Black> he referred to them as different primitives
[14:33:35] <Robert Cragie> Yes
[14:34:04] <Robert Cragie> There is a clue in the name "response"
[14:34:16] <Robert Cragie> It is a response - it doesn't solicit another response
[14:34:45] <Robert Cragie> Trying to overload the underlying transport to imply some custody transfer doesn't make a lot of sense to me
[14:34:59] Lucas Jenss leaves the room
[14:35:37] <Robert Cragie> Surely that is the important point - does the request or response get delivered reliably?
[14:35:57] <Robert Cragie> Whether that is done by CoAP messaging layer or transport layer?
[14:40:32] <Benjamin Black> [mic] a notification from an observe doesn't have a response!
[14:40:51] <Benjamin Black> [mic] how do i know it has gotten to the app buffer on the receiver?
[14:40:51] Rosen, Brian leaves the room
[14:41:03] Rosen, Brian joins the room
[14:43:00] <Benjamin Black> i was responding to simon asserting a response is sufficient.
[14:43:16] Lucas Jenss joins the room
[14:43:51] M Jenkins leaves the room
[14:43:55] <Robert Cragie> OK sorry for the bad sync.
[14:44:12] <Benjamin Black> no worries, amazing it is working even this well :)
[14:51:48] Paventhan Arumugam leaves the room
[14:56:17] <Benjamin Black> have to drop off, thanks for the help, robert
[14:56:21] Benjamin Black leaves the room
[15:03:44] HS YANG leaves the room
[15:14:41] mmani leaves the room
[15:19:04] Thomas Watteyne leaves the room
[15:21:59] <Robert Cragie> Meeting is now over
[15:22:16] kaorumaeda leaves the room
[15:22:17] Robert Cragie leaves the room
[15:22:31] Barry Leiba leaves the room
[15:22:37] Rosen, Brian leaves the room
[15:22:44] Meetecho leaves the room
[15:23:55] pascal thubert leaves the room
[15:29:47] akeranen leaves the room
[15:33:50] kaorumaeda joins the room
[15:37:47] Lucas Jenss leaves the room
[15:39:16] Lucas Jenss joins the room
[15:39:30] Barry Leiba joins the room
[15:41:44] Barry Leiba leaves the room
[15:46:31] renzoe leaves the room
[15:54:29] kaorumaeda leaves the room
[15:57:40] Rosen, Brian joins the room
[16:21:07] Lucas Jenss leaves the room
[16:37:07] Rosen, Brian leaves the room
[16:40:01] Lucas Jenss joins the room
[17:03:55] Lucas Jenss leaves the room
[20:39:52] Rosen, Brian joins the room
[20:44:10] Rosen, Brian leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!