16:09:10 #startmeeting prop224 implementation plan 16:09:10 Meeting started Mon May 30 16:09:10 2016 UTC. The chair is asn. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:09:10 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:09:12 ah nice 16:09:17 aight 16:09:43 OK so I made some notes of the things I consider TODO short-term wrt prop224 16:10:09 go 16:10:31 OK, so there are both spec and code things that we need to do on the short-term 16:10:38 let me start by spec, which is shorter: 16:10:44 IMO we need to do the following ASAP: 16:10:57 - Finalize descriptor stuff (AFAIK almost ready but special has some fixups) 16:11:12 - Finalize client auth stuff (not sure where we are here, but we need to pin this down RSN) 16:11:40 - Finalize time period stuff (this is not super urgent, but there are thigns we discussed in montreal that have not been written down in the spec. and I need to do ASAP before I forget things.) 16:11:47 these are the three spec things I consider "urgent" 16:11:58 mainly because the implementation is blocking 16:12:15 EOF for now 16:12:20 any comments? 16:12:45 sounds good to me and agree on the urgent 16:12:50 aight 16:12:53 mm 16:12:56 we have most of them finalized except maybe client auth I think? 16:13:01 right 16:13:02 yes 16:13:11 client auth might require some discussion back & forth 16:13:48 ok 16:13:52 so that's that on the spec side 16:13:57 i guess you don't have anything else that is urgent 16:14:08 I suspect there are also some cell related changes that will become "urgent" once we start working on implementing those cells 16:14:14 special: agreed! 16:14:47 unfortunately, we will probably get additional insight on things when we start implementing. 16:14:55 yah 16:15:05 so let me proceed with the code tasks that I consider short-term TODO: 16:15:18 (we should probably put these lists on a pad later on this meeting) 16:15:42 - Refactor rend_data_t so that we can do descriptor upload/fetching from HSDirs 16:15:53 -- Finalize HSDir-related code (depends on above) 16:16:26 - Start implementing cells (maybe start from ESTABLISH_INTRO aka #19043) 16:16:46 - Finalize prop250 stuff (take care of review, and the SR state open bugs) 16:17:05 These are the coding tasks I think are relevant for the next days/weeks. 16:17:23 agree, 250 and rend_data_t are my priorities 16:17:28 - Finalize descriptor code (& all of the tests for it) 16:17:29 I also had a note about asking teor whether he can spend some time to think how we can make chutney more useful for future prop224 testing 16:17:34 unless that was part of "HSDir code" 16:17:43 special: yes that's what i meant 16:17:52 I think they are different things 16:17:54 yeah 16:17:55 ah 16:17:59 you probably know better 16:18:09 encode/decode, hsdir cache, hsdir upload/download basically 16:18:19 the later is the one we need rend_data_t 16:18:34 #17238 16:18:38 see child tickets 16:18:45 we could do better breakdown ^ 16:19:16 hm ok 16:19:18 let me launch a pad 16:19:21 maybe it will help us 16:19:22 coordinate this 16:19:44 asn: we do have a pad for 224 HSDir btw 16:20:18 ok let's use this then: https://storm.torproject.org/shared/HoK_R0_ohIZa4e2Ef52vrhfiVC1dzDD658vyK14bCV2 16:20:31 i think this is the link to the prop224 pad [omg leak] 16:20:53 that one is on tor-dev@ somewhere I think unless your link is RW rights :) 16:21:12 ok let's go to the bot 16:21:15 bottom 16:23:13 ok 16:23:15 what is the extra indentation meant to imply 16:23:31 i was trying to imply dependencies 16:23:50 in the sense that rend_data_t refactoring is required to get the HSDir thing together 16:24:01 i think 16:24:10 but maybe that's not useful data 16:24:26 it might be, just wasn't appropriate for the descriptor part. it's better now. 16:24:36 :) 16:25:05 OK 16:25:06 ok there some ticket numbers 16:25:09 thx 16:25:14 do you guys feel we should add more stuff in there? 16:25:26 i think if we split those tasks between us, we will be busy for N days 16:25:29 if this is for the short-term, it's good I think 16:25:31 only for* 16:25:44 where N is probably greater than 5 and less than 25 16:25:45 * armadev creates #19204 from one of the montreal todos 16:25:46 if we were to add more stuff, it'd be "more cells"? 16:26:25 armadev: I was wondering the other day, why exactly this is needed for 224 ? 16:26:48 #19024 and #19204 are both related, and anagrams! 16:27:53 dgoulet: i think it isn't. except if we want to change stuff that uses rend_service_descriptor_t, we're going to have an easier time of it if the struct is easier to use. 16:28:02 right ok 16:28:14 i'm just following through with a plausible idea that nick had :) 16:29:25 if we think that ticket is worth doing, we should make it a child of another ticket, otherwise it will be lost in the void 16:29:35 so I've already started a bit at the meeting the rend_data_t part so answering nickm's review of 250 and rend_data_t is what I can do in the short term which I can transition in finalizing the fetch/store logic of HSDir 16:30:10 special: seems you have the encode/decode under control? or not? 16:30:15 yes 16:30:18 i'd also be interested in doing some prop250. i feel that the design has drifted from my brain during the past weeks, and I'd like to review it again. 16:30:37 so you guys agree we should try to assign names to these tasks? 16:30:42 i will put my name in the obvious parts 16:30:43 yes 16:30:45 yeah 16:30:54 asn: if we want it to be a child, it could be the child of one of the "publish new rend descs to new hsdirs" tasks? 16:31:15 I'm the owner of #19024 so I guess my name is on it 16:31:36 I'll make a child ticket for "HSDir fetch/store logic" 16:32:01 dgoulet: can you also take armadev's ticket and place it somewhere in your HSDir ticket infrastructure? 16:32:16 i can also take over the ESTABLISH_INTRO thing if you guys are fine with it 16:32:23 totally 16:32:40 asn: hrm... #19204 is kind of not 224 but if it's tagged tor-hs, we'll not forget about it :) 16:32:49 i'd love to do some code review. but what i want most is to not be presented with a 6000 line patch that mostly just moves stuff around but also made a couple of simple changes at the same time. 16:33:06 i looked at one of the branches and was immediately faced with such a 6000 line patch. i couldn't proceed. 16:33:13 one of what branches? 16:33:25 i don't know. something in montreal. 16:33:34 not sure we have a branch that big and complicated for 224, I think you looked at the prop250 branch :) 16:33:39 armadev: and yes it's big 16:33:45 plausible! 16:34:41 ok i think we have names attached to the crucial tickets in there 16:34:50 s/tickets/tasks/ 16:35:06 personally, i plan to start with prop250. then move to spec. then move to cells. 16:35:09 there added mine for the subtask of finalizing because that's currently mostly my code 16:35:31 we can touch base again in a week or two. we are all pretty active on IRC, so we can communicate on this as the days go by? 16:35:38 mhm 16:35:43 asn: I can help you with time period if needed, it's still fresh in my memory 16:35:44 I have some distractions happening this week 16:36:03 dgoulet: ok, i will make a patch and then you can review it? 16:36:07 perfect 16:36:24 that's why I added my name to the other spec patches as well. because I plan to spend some time replying to special's client auth thread, etc. 16:36:27 am very excited to read the time period document with a fresh mind and see if it makes sense 16:36:28 is there a page that lists the most important needs-review 224 and 250 branches/patches/commits? :) 16:36:54 armadev: query with "prop250" and "prop224" in the Summary 16:37:17 armadev: they both have super meta top parent ticket, #16943 and #12424 16:37:54 armadev: and also in the pad asn shared above :) 16:38:27 armadev: i think we will have a final final prop250 branch in the upcoming week. you might want to hold your review moments for some of that. 16:38:40 big words there, but let's see. 16:38:50 "final final" you mean nickm approved? 16:39:20 hmm maybe not "ready to merge by nickm", but more like "asn and dgoulet and testnet are all happy with it" 16:39:41 ok 16:39:51 great. carry on, let me know if you run across a thing this week that i would be especially helpful for. 16:39:58 my schedule gets ugly again on saturday. 16:40:46 * armadev proceeds to open those other tickets he said he would open based on nickm's suggestions 16:40:51 ack 16:41:17 dgoulet: special: ok are we happy? 16:41:20 I am 16:41:24 yes 16:41:40 great 16:41:43 #endmeeting