[00:24:13] blassey joins the room
[12:05:37] lucasp joins the room
[12:06:01] mnot joins the room
[12:06:19] mnot leaves the room
[12:06:28] mnot joins the room
[12:45:52] blassey leaves the room: Disconnected: closed
[12:45:52] lucasp leaves the room
[12:47:14] blassey joins the room
[12:47:14] blassey leaves the room: Disconnected: Broken pipe
[12:59:41] dschinazi@jab.im joins the room
[12:59:46] Mike Bishop joins the room
[13:01:24] blassey joins the room
[13:02:25] James Gruessing joins the room
[13:03:06] lucasp joins the room
[13:03:30] <Mike Bishop> Codimd is saying I'm offline, so no contribution to minutes for me.
[13:03:38] afrind joins the room
[13:05:34] ioggstream joins the room
[13:21:29] <lucasp> https://github.com/martinthomson/http2-spec/issues/19 for hpack. Ian's suggestion of a new ALPN would support changing things in a breaking way. I'm not going to push hard for a HPACK change
[13:21:30] James Gruessing leaves the room
[13:21:30] lucasp leaves the room
[13:21:35] Jeffrey Yasskin joins the room
[13:23:09] lpardue2 joins the room
[13:34:27] <Jeffrey Yasskin> Unlike most other GREASE mechanisms I've seen, which explicitly reserve the GREASE codepoints, this document says that it's important not to say which field names are reserved. What's up with that?
[13:34:47] httpbis@jabber.ietf.org joins the room
[13:36:35] <Jeffrey Yasskin> Thanks mnot!
[13:36:36] httpbis@jabber.ietf.org leaves the room
[13:37:28] lucasp joins the room
[13:37:29] lucasp leaves the room
[13:37:33] lnicco@jabb.im joins the room
[13:38:03] <Jeffrey Yasskin> *coughenterprisecough*
[13:41:24] <mt> Those people who are unhappy with too many headers are going to be unhappy.
[13:41:38] <mt> I don't think that this needs to be published at all.
[13:49:09] <mt> This isn't really a problem you can solve by publishing specs.
[13:50:01] <mt> This is community building.  We build a community and determine who is part of that community by virtue of what we tolerate and how we comport ourselves.
[14:01:39] <mt> 0.5-RTT works perfectly well.
[14:07:02] <mnot> I'm starting to wonder how many CH headers there will be (that are actually used)
[14:07:38] <mt> Ah, I don't think that Mike's thing is going to work: ALPN/AltSvc is authenticated by the handshake.
[14:08:12] <mt> Very few from Firefox.  We have looked at precisely one CH (I think that it was DPR).
[14:08:16] <mnot> (in other words, can we shove this into a SETTINGS value as a bitmask? :)
[14:10:20] <mt> Ahh, that's an interesting approach: you can get a lot of headers from a single setting.
[14:10:45] <mnot> enough, with a little registry. It's ugly, but I wonder if it's as ugly as this might be.
[14:12:42] <mt> This DNS thing doesn't really help for anything you might want to authenticate, as it turns out.  You have to do the TLS handshake thing as well.
[14:13:03] <Mike Bishop> David, I think you're making a circular argument there.  DNS isn't an optimization because it gives you less than other solutions we also haven't built yet.
[14:14:43] <mt> Chairs, the pacing of these discussions has been pretty good.  The risk that we have a gong show is strong.
[14:14:57] <mt> This has been about the right balance.
[14:15:15] <Mike Bishop> Can someone remind what's going on with Variants?  It appears to be expired, but not marked as Dead.
[14:15:48] <mt> I still kinda like variants; it probably needs a lot more implementationlove though.
[14:16:20] <mnot> Mike: I was waiting for Structured Headers to ship.
[14:16:50] <mnot> … and then I started to think about a slightly different approach, which I haven't had a chance to write down as a proposal yet.
[14:17:08] <mnot> So it's Hibernating with Potential, I suppose.
[14:19:18] <mt> Implementation interest?
[14:21:53] <mnot> I do wonder if Stack Overflow will realise that they can make us do things if we adopt this...
[14:22:39] <mt> Oh, they already know that.  There is a control surface in both directions, equally weak.
[14:23:59] <ioggstream> +1 for SEARCH
[14:24:58] <Mike Bishop> RFC 8470 keys off of safe, not idempotency.  This is still safe.
[14:25:46] <mt> I know that we used safe, but idempotency is something you can use to make that decision.  But it's far riskier.
[14:25:58] <ioggstream> imho once we standardize it, they will come :) having a safe way to search is  required
[14:26:04] <mt> Individually idempotent operations still have effects.
[14:28:38] <mt> I did briefly consider the potential for a mechanism to indicate that "POST to this resource is safe", but that doesn't work if the server doesn't ever get to speak.
[14:30:21] afrind leaves the room
[14:30:25] Mike Bishop leaves the room
[14:31:43] mnot leaves the room
[14:32:57] lnicco@jabb.im leaves the room
[14:36:27] Jeffrey Yasskin leaves the room
[14:52:38] lpardue2 leaves the room: Disconnected: BOSH client silent for over 60 seconds
[15:00:29] lucasp joins the room
[15:08:00] dschinazi@jab.im leaves the room
[15:52:48] Matthew[m] joins the room
[17:03:44] ioggstream leaves the room
[17:08:57] ilari.liusvaara joins the room
[17:10:13] ilari.liusvaara leaves the room
[17:10:22] ilari.liusvaara joins the room
[17:10:40] ilari.liusvaara leaves the room
[18:01:12] James Gruessing joins the room
[18:07:47] James Gruessing leaves the room
[18:19:37] James Gruessing joins the room
[18:21:38] James Gruessing leaves the room
[18:31:22] blassey leaves the room
[18:36:56] blassey joins the room
[19:01:51] lucasp leaves the room
[19:02:53] lucasp joins the room
[19:32:34] chi.jiun.su joins the room
[20:36:21] blassey leaves the room
[21:28:48] lucasp leaves the room
[21:28:50] lucasp joins the room
[21:28:50] lucasp leaves the room
[21:33:57] lucasp joins the room
[22:22:57] chi.jiun.su leaves the room
[22:38:34] lucasp leaves the room
[22:48:39] lucasp joins the room
[22:50:59] lucasp leaves the room
[22:51:15] lucasp joins the room
[23:12:37] lucasp leaves the room
[23:28:15] James Gruessing joins the room
[23:30:11] James Gruessing leaves the room
[23:34:14] bridgebot joins the room