[00:02:29] Questioner: I strongly support your work [00:02:41] Q: average diameter, ver per link, packet error lrate [00:02:46] Q: useful piece of work [00:02:52] Q is actually JP [00:02:58] JP: not need to cover ECN in same document [00:03:05] JP: different drafts you plan? [00:03:15] PT: if the groups could work on more I'd like to [00:03:35] PT: when you get this, go back to source... the source is the one controlling the ack... no congestion, the last fragment [00:03:43] PT: if congestion it should more ... ack [00:03:56] JP: thinking of more generic, way to use it for recovery would be in the other draft. [00:04:10] HP: flow control-should be made light enough, don't invent TCP [00:04:16] PT: right, just take tcp layer... [00:04:22] HP was JP [00:04:25] PT: not the slowstart part [00:04:42] PT: is this relevant work for us? work list ours? [00:05:04] Carsten: as personal oppinion, this is essentially doing ed transport to the link-layer [00:05:10] PT: like vlc 802. [00:05:27] Carsten: subnet advice document, 38-something, we wrote couple of years ago [00:05:34] Carsten Borman is CB [00:05:41] CB: we defined situations like this [00:05:53] CB: there are certainly reasons to do somehting like this, when considerable optimizations... [00:06:12] CB: there are also adonitions in the document, where you consider iterations at the link layer, what would the transport do [00:06:31] CB: if you run TCP on this, then interesting, because TCP has own control loops, here interacting to there [00:06:45] PT: what we do here, within TCP segmnetn, slower clock on the link layer. [00:06:52] PT: not a feedback loop on the tcp [00:06:56] PT: on the resolution side. [00:07:08] CB: exampleI give: you have some form of asking an ack at link layer [00:07:18] CB: you probably ask for this after the fragments arrived [00:07:26] CB: you adding some delay [00:07:30] CB: retransmitting appears [00:07:37] CB: until time has been transmitted [00:07:44] CB: time measured for previous packets [00:07:51] CB: you have look at TCP RTO calculations [00:07:57] PT: draft, I wrote that a year ago [00:08:00] PT: below [00:08:03] below Tcp [00:08:17] CB: 3819 I think it is, could someonconfirm [00:08:24] CB: look at that, don't fall in the same traps [00:08:32] JP: true for other mechanisms as well [00:08:36] JP: IP over optical [00:08:53] JP: recovery at one layer and at another layer, weird,; better maybe add a section here [00:09:09] JP: that might be the way to do it, implementer tweaks mechanism [00:09:14] JP: back to your example of tcp [00:09:18] JP: low packets [00:09:37] JP: if you don't have these mechanisms... no exponential backoff... especially tcp... [00:09:44] what would be a good MSS? [00:09:52] JP: that could be part of the draft [00:10:02] JP: that's righ. [00:10:03] right. [00:10:23] I can't hear. [00:10:29] JP: thanks [00:10:50] I still can't hear a questioner, like s/he's very far away - does s/he use mike? [00:11:38] CB: wolf actually thinks doing retransmissions... [00:11:47] CB: JP already said that [00:12:22] Questioner: academic discussion coul dbe about fragment recovery, UDP, the tag, size, offset. Do recovery without bitmpa. I'm here to stresss we need an interoperable way. [00:12:30] Questioner (ISA): ISA needs this. [00:13:16] JP: I supported this work as long as PT agrees: how to add... [00:13:23] PT seems to be speaking but can't hear. [00:13:49] Jonathan Hui is JH [00:14:02] JH: when radio hops, fragments can traverse multiple hops - clear [00:14:22] JH: less clear, organizationa... a datagram hop by hop, you want essenetially, recovery on a single hop. [00:14:33] PT: speaks... can't hear. [00:15:25] CB: the argument o fJonathan is already we have at MAC layer ack [00:15:35] CB: whereas your argument is the frags could go different paths [00:15:39] PT: speaks. [00:15:49] CB: we handle a big canyon here, how big is the target we aim at? [00:15:56] CB: We need to understand the reqs. [00:16:02] CB: some for of validation I want to see. [00:16:09] CB: simulations some forms of. [00:16:15] CB: another [00:16:20] PT: speaks. [00:17:30] CB: wg interested in this work, show hands, implementing? [00:17:36] CB: a pretty good turnout of hands [00:17:52] CB: may have looked small, but in the room there aren't many people, sounds interesting. [00:18:05] CB: enjoy cookies. [00:18:15] Sounds as it's adjourned. [00:20:15] james.dishongh@gmail.com joins the room [00:20:45] james.dishongh@gmail.com leaves the room [00:21:21] bigbanana joins the room [00:21:21] Hiroki leaves the room [00:21:21] Pascal Thubert leaves the room [00:21:36] yuichi igarashi leaves the room [00:22:11] Bluesheets anyone? [00:22:26] Alex Petrescu leaves the room [00:25:58] bigbanana leaves the room [00:30:26] Alex Petrescu joins the room [00:35:53] sdhags leaves the room [00:41:28] Shaun Kaplan leaves the room [00:42:19] ShoichiSakane leaves the room [00:49:57] Hiroki joins the room [00:50:30] Hiroki leaves the room [01:01:56] antonio.moreiras leaves the room [01:15:48] Alex Petrescu leaves the room: Replaced by new connection [14:49:08] fmcway joins the room [14:49:14] fmcway leaves the room