httpbis - Thursday, June 5, 2014

httpbis

httpbis@jabber.ietf.org

Thursday, June 5, 2014< ^ >

Julian has set the subject to: https://github.com/http2/wg_materials/blob/master/interim-14-01/agenda.md

Room Configuration

Room Occupants

GMT+0
[13:11:48] mnot joins the room
[13:11:50] <mnot> hello
[13:12:16] Julian joins the room
[13:12:39] mcmanus joins the room
[13:12:43] Barry Leiba joins the room
[13:14:21] Tony Hansen joins the room
[13:15:22] <mcmanus> https://etherpad.mozilla.org/httbis-interim
[13:16:56] <mcmanus> https://etherpad.mozilla.org/httpbis-interim <https://etherpad.mozilla.org/httbis-interim> (typo fix)
[13:27:41] roberto_peon joins the room
[13:28:25] <Barry Leiba> Hi, Roberto.
[13:28:30] <roberto_peon> Hello
[13:28:45] mnot waves
[13:28:58] <roberto_peon> Is there a way to listen in yet?
[13:29:08] <mnot> no
[13:38:57] <Barry Leiba> Roberto, want to Skype with me? You should be able to hear most of what's happening.
[13:39:12] <mnot> we can try to set up a dial in on the break
[13:39:35] <roberto_peon> cool.
[13:39:44] <Barry Leiba> Can use Skype for now, as a stop-gap, if you want.
[13:39:45] <roberto_peon> For now, I'm listening in via Will's computer.
[13:39:48] <Barry Leiba> OK
[13:41:09] Julian leaves the room: Replaced by new connection
[13:41:10] Julian joins the room
[13:41:54] <roberto_peon> alt-svc can't be done as a header unless there is an outsstanding request.
[13:42:29] <roberto_peon> This is true of any feature that one would want to do where there isn't a request.
[13:49:44] <roberto_peon> anyone out there to relay?
[13:49:56] <mnot> i will
[13:50:46] <roberto_peon> Thx: 'hont'-style extensions should not need negotiation before use (else they're not usable for an RT minimum), they should be ignored, though a mechanisms to say 'stop' should be there (which could be the negotiation).
[13:50:59] <roberto_peon> Extensions that change protocol semantics, of course, must need the negotiation.
[13:51:02] <mnot> hont?
[13:51:07] <roberto_peon> s/hont/hint
[13:51:46] <roberto_peon> and by 'they should be ignored', I mean that, if unknown, then ignored.
[13:55:33] Hervé joins the room
[13:57:14] <roberto_peon> Yes please on the h2-e ALPN token
[14:02:09] <roberto_peon> agreed, that is my understanding as well.
[14:05:48] <roberto_peon> DNS-push shouldn't require a roundtrip, as it is a hint-style thing.
[14:19:18] ilari.liusvaara joins the room
[14:23:22] <roberto_peon> imho: hint-based (ignorable) extensions for "h2", and the rest on ALPN "h2-e", which is http2 plus extensions (to be defined in another draft)
[14:27:23] <roberto_peon> Can someone relat what I said above?
[14:27:27] <roberto_peon> *relay
[14:33:48] <roberto_peon> the base spec shouldn't be gated on extensibility.
[14:34:01] <roberto_peon> hint-based extensibility shouldn't be blocked on state-changing extensibility.
[14:34:55] <roberto_peon> Either is fine.
[14:43:59] <mcmanus> btw daniel's scribing is amazing
[14:44:41] <roberto_peon> yup
[14:49:44] <roberto_peon> don't need the frame ID even.
[14:49:49] <roberto_peon> just don't send any extensions.
[14:50:18] <roberto_peon> I_AM_A_STRICT_TASKMASTER seting :)
[15:23:12] <roberto_peon> Without the setting, you end up with not being able to distinguish between de-synchronized and other stuff, at least theoretically.
[15:23:29] <roberto_peon> standing up
[15:47:06] <roberto_peon> The only thing we need to enable WS is to finish the discussion on headers and flow control.
[15:47:15] <roberto_peon> we need to finish that discussion regardless.
[15:47:19] <roberto_peon> please relay :)
[15:48:29] <roberto_peon> jhungry here:)
[15:49:55] Julian leaves the room: Computer went to sleep
[15:50:51] Hervé leaves the room
[15:50:55] Hervé joins the room
[15:56:05] Hervé leaves the room
[15:56:17] Hervé joins the room
[16:04:29] mnot leaves the room
[16:15:30] Hervé leaves the room
[16:22:33] Julian joins the room
[16:29:19] roberto_peon joins the room
[16:33:14] mnot joins the room
[16:36:26] Hervé joins the room
[16:45:21] jeffmendoza joins the room
[16:45:48] <roberto_peon> Note that this can entirely be a server-based decision, and thus application-based.
[16:46:12] <roberto_peon> The issue is the gateway problem between the http/1.1 server and the http/2 server
[16:46:26] <roberto_peon> yes
[16:46:26] <roberto_peon> keep it
[16:47:18] jeffmendoza leaves the room
[16:57:59] jeffmendoza joins the room
[16:59:03] jeffmendoza joins the room
[16:59:18] jeffmendoza leaves the room
[17:01:17] jeffmendoza joins the room
[17:01:18] jeffmendoza leaves the room
[17:03:18] jeffmendoza leaves the room
[17:03:20] jeffmendoza joins the room
[17:07:18] jeffmendoza leaves the room
[17:07:34] jeffmendoza joins the room
[17:08:48] jeffmendoza leaves the room
[17:36:00] Barry Leiba joins the room
[17:46:24] Barry Leiba leaves the room
[18:12:48] roberto_peon leaves the room
[18:41:52] Hervé leaves the room
[18:44:46] Tony Hansen leaves the room
[18:59:17] roberto_peon leaves the room
[19:01:59] Hervé joins the room
[19:04:17] Tony Hansen joins the room
[19:08:16] Julian leaves the room: Replaced by new connection
[19:08:16] Julian joins the room
[20:05:43] Tony Hansen joins the room
[20:06:59] Tony Hansen leaves the room
[20:22:06] Barry Leiba leaves the room
[20:22:56] Barry Leiba joins the room
[20:56:46] Tony Hansen leaves the room
[20:57:31] Tony Hansen joins the room
[21:06:16] Barry Leiba leaves the room
[21:06:23] Tony Hansen leaves the room
[21:06:25] mnot leaves the room
[21:06:27] Julian leaves the room
[21:06:33] Hervé leaves the room
[21:06:47] Hervé joins the room
[21:07:28] Hervé leaves the room
[21:13:53] mcmanus leaves the room
[21:51:00] mcmanus joins the room
[23:47:57] mcmanus leaves the room