00:51:12  <Domenic>littledan: do you have a writeup for static blocks somewhere?
00:53:07  * Fishrock123joined
00:58:50  * Fishrock123quit (Remote host closed the connection)
00:59:30  * Fishrock123joined
00:59:38  * Fishrock123quit (Remote host closed the connection)
01:31:35  <annevk>aklein: but you effectively want to have run all of test262 in each global as well, duplicating all those tests seems like a waste of effort
02:08:27  * Fishrock123joined
02:13:02  * Fishrock123quit (Ping timeout: 260 seconds)
03:07:07  <cloudshu>annevk: you mean intra agent tests?
03:07:07  <cloudshu>annevk: those should remain in test262
03:09:20  <annevk>cloudshu: I meant that any given test should be run in each global
03:10:31  <annevk>cloudshu: for intra-agent tests it would similarly make sense if they can be run (and are) in a wide variety of host setups
03:12:47  <cloudshu>annevk: sure but i'm not understanding the duplication concern
03:14:07  <annevk>cloudshu: are you suggesting WPT should wrap test262 in this way?
03:14:45  <cloudshu>annevk: give me 5 to get to a computer
03:16:35  <dilijev>bterlson: in the tc39 meeting notes it says you mentioned having a checkbox to inline annexB within the spec webpage, on the surface, I think that would be a great compromise that would help bring that information into easy view for implementers, regardless of whether AnnexB is ultimately inlined
03:17:28  <bterlson>dilijev: yeah, unfortunately requires significantly more tooling work :-P
03:29:57  <cloudshu>annevk: back. let's make sure we're talking about the same things
03:30:41  <cloudshu>annevk: are we talking about additional agent tests, or are we talking about how you want to run all of test262 in various different web globals, which currently nobody does in practice?
03:32:01  <annevk>cloudshu: I've been talking about both those things
03:35:04  <cloudshu>annevk: running test262 in different web globals is up to the different vendors right now, since the test262 test infra abstracts over host-provided stuff, or attempts to
03:36:00  <cloudshu>annevk: the fact it doesn't happen in practice, like Domenic said...
03:36:22  <cloudshu>annevk: i guess we could encourage the various implementers to try to do this, but i'm not sure how realistic that'll be
03:36:44  <cloudshu>annevk: the additional agent tests on the other hand, have two flavors of tests:
03:36:56  <cloudshu>annevk: 1) stuff that tests behavior given an agent with some configuration
03:37:12  <cloudshu>annevk: 2) stuff that tests a web global has a certain configuration
03:37:44  <cloudshu>annevk: 1) can live in test262 and would require some thinking about how to actually abstract over "run this in an agent with configuration C", which doesn't exist right now
03:37:51  <cloudshu>annevk: 2) should go into WPT
04:45:01  * jmdyckquit (Quit: Leaving.)
04:47:17  * gskachkov_joined
04:52:11  * gskachkov_quit (Quit: gskachkov_)
05:10:09  * Fishrock123joined
05:15:12  * Fishrock123quit (Ping timeout: 245 seconds)
05:23:45  <annevk>cloudshu: apologies, got distracted with GitHub and breakfast
05:24:02  <cloudshu>annevk: no worries, was having GH and dinner myself
05:24:19  <annevk>cloudshu: it seems to me that 1/2 are very much intertwined
05:24:57  <annevk>cloudshu: once you have tests for each agent config, you can reuse that for host agents
05:26:32  <cloudshu>annevk: do you want something like test262 to test that, e.g., ServiceWorker agents conform to some configuration?
05:28:12  <annevk>cloudshu: dunno, but I don't want to duplicate tests
05:30:44  * Fishrock123joined
05:30:44  <cloudshu>annevk: i don't think there's gonna be duplication
05:30:54  <cloudshu>annevk: i'm thinking of something like,
05:30:58  <annevk>cloudshu: I could imagine some setup where you have tests for different agent profiles that are generic and we load them in specific host agents
05:31:09  <cloudshu>annevk: yes, that's what i'm thinking
05:31:31  <cloudshu>annevk: it so happens that the profiles match the different worker configurations we care about on the web
05:31:57  <annevk>Okay, so it seems that requires some coupling if you feel the host agents should be in WPT
05:32:03  <cloudshu>annevk: but as a policy point test262 isn't going to directly test those workers
05:32:29  <annevk>Or duplication
05:32:41  <cloudshu>annevk: no, what i feel should be in WPT is the test that ServiceWorker corresponds exactly to some profile
05:32:46  <cloudshu>annevk: i guess what'll be duplicated is the info about those profiles
05:32:49  <cloudshu>annevk: not the tests themselves
05:32:55  * Fishrock123quit (Client Quit)
05:32:56  <cloudshu>ServiceWorker or whatever worker
05:33:26  <annevk>Isn't that the same test as an agent profile test?
05:33:56  <annevk>As far as I can there's only two variants, CanBlock or not
05:34:02  <cloudshu>annevk: no -- the agent profile test tests that the observable behavior via JS execution is correct
05:34:07  <cloudshu>annevk: to be concrete about CanBlock,
05:34:29  <cloudshu>annevk: the agent profile test would be a test that asserts Atomics.wait throws when the agent profile is [[CanBlock]] == true, and does not throw when false
05:34:44  <cloudshu>annevk: the WPT test would be ServiceWorker has a certain [[CanBlock]] value (i don't know what it is, actually)
05:35:40  <annevk>We'd also want to run the profile test though, to make sure there are no bugs
05:36:06  <annevk>In case UAs make agent-specific optimizations or code paths
05:36:15  <cloudshu>annevk: you mean to run it in a real ServiceWorker?
05:36:20  <annevk>Yes
05:36:30  <cloudshu>annevk: point taken, but i don't think test262, as a matter of policy, is going to mandate that
05:36:40  <cloudshu>i personally would like to see test262 own up more to the web nature of JS but
05:36:50  <cloudshu>i'm not doing the maintenance ;)
05:37:04  <annevk>Okay, but that was my point
05:37:21  <annevk>That you either need coupling or duplication
05:37:44  <cloudshu>annevk: maybe i don't understand what you mean by duplication
05:37:49  <annevk>Or some hope that stuff is invariant across agents/globals, which is meh
05:38:07  <cloudshu>annevk: i think you just hope that we, Chrome, Safari, and Edge all run those agent profile tests under the actual workers
05:38:15  <cloudshu>instead of some simulated shell agent
05:38:39  <annevk>cloudshu: yeah, and all types too
05:39:05  <cloudshu>annevk: this is analogous to a current problem with running it in the shell or in a browser env
05:39:17  <annevk>cloudshu: with worklets we now have 5 agent types, and 6 diff globals or some such
05:39:26  <cloudshu>annevk: i can't speak to other vendors' CIs, but we run it in both right now,
05:39:48  <cloudshu>annevk: and when i first enabled it, a bunch of tests failed because the tests were written without considering HTML top-level names that were getters/setters
05:42:01  <cloudshu>annevk: i'll bring this up in committee next month
05:42:09  <cloudshu>annevk: it's really just the problem that test262 ignores web reality
05:43:45  <cloudshu>and while thus far it's been a matter of annex b and some special top-level names, it's gonna get more real with SAB
05:43:56  <cloudshu>my fear, though, is that it's just gonna get swept under the rug as "not our problem"
05:44:00  <cloudshu>let the vendors deal with it
05:51:38  <annevk>cloudshu: one way we could deal with it is import the whole thing in WPT somehow with some instrumentation on top to make sure all combinations are covered
05:52:06  <cloudshu>annevk: that'd be neat
05:52:07  <annevk>cloudshu: some kind of build script that creates a WPT wrapper for all the various host combinations
05:52:32  <cloudshu>annevk: PM
06:08:33  * brabjoined
06:24:11  * gskachkov_joined
07:00:20  * gskachkov_quit (Quit: gskachkov_)
07:01:34  * gskachkov_joined
07:23:45  <dilijev>random question about the tc39 meeting notes. They seem very detailed with quotes (or summaries) of what each attendee had to say. I'm wondering how close to a transcript is this or is it suitably summarized? I ask because in places it seems like a transcript, but I find it hard to believe that the note takers are truly attempting to make a verbatim
07:23:45  <dilijev>transcript of the meeting. I imagine a true transcript would also not be as useful as a distilled summary of points.
07:27:49  <ljharb>it's definitely not always a detailed transcript. i think you imagine correctly.
07:28:24  <littledan>dilijev: Recently it has been more like a transcript, so it can be hard to get the points. The last couple meetings had summary docs written. Maybe we need a third, intermediate form
07:28:39  <littledan>an issue is that it can be hard for note takers to extract correct points from everyone
07:29:04  <littledan>Domenic: https://github.com/littledan/proposal-class-static-block
07:53:00  <annevk>littledan: curious, were recordings ever considered?
07:53:34  <cloudshu>littledan: ping
07:56:54  <cloudshu>littledan: nvm i'll just comment in issue, heading to bed
08:14:00  <ljharb>annevk: if we had them, they'd be only useful for creating edited transcripts; i don't think we'd all feel comfortable speaking freely if every word was publicly available.
08:38:53  * not-an-aardvarkquit (Quit: Connection closed for inactivity)
08:53:55  * gskachkov_quit (Quit: gskachkov_)
08:55:15  * gskachkov_joined
09:01:38  * gskachkov_quit (Quit: gskachkov_)
09:16:49  * gskachkov_joined
09:17:48  * gskachkov_quit (Client Quit)
09:19:07  * gskachkov_joined
09:20:03  * gskachkov_quit (Client Quit)
09:20:57  * gskachkov_joined
09:25:48  * gskachkov_quit (Client Quit)
10:25:09  * mylesborinsquit (Quit: farewell for now)
10:25:39  * mylesborinsjoined
11:34:34  * jmdyckjoined
13:34:33  * bradleymeckjoined
14:14:29  * Fishrock123joined
14:27:44  * bradleymeckquit (Quit: bradleymeck)
14:55:10  * brabquit (Ping timeout: 240 seconds)
15:03:40  * mmun_joined
15:03:52  * caiolima__joined
15:04:27  * brianloveswords_joined
15:05:38  * tobie_joined
15:06:33  * bradleymeckjoined
15:07:29  * cloudshu_joined
15:10:49  * dilijevquit (*.net *.split)
15:10:50  * brianloveswordsquit (*.net *.split)
15:10:51  * tobiequit (*.net *.split)
15:10:51  * mmunquit (*.net *.split)
15:10:51  * gsathyaquit (*.net *.split)
15:10:53  * caiolima_quit (*.net *.split)
15:10:54  * cloudshuquit (*.net *.split)
15:10:54  * akleinquit (*.net *.split)
15:10:55  * cloudshu_changed nick to cloudshu
15:10:58  * brianloveswords_changed nick to brianloveswords
15:10:58  * mmun_changed nick to mmun
15:11:40  * gsathyajoined
15:12:14  * tobie_changed nick to tobie
15:14:24  * dilijevjoined
15:37:36  * surmaquit
15:37:51  * surmajoined
16:32:04  * mmun_joined
16:32:06  * brianloveswordsquit (Ping timeout: 252 seconds)
16:32:06  * mmunquit (Ping timeout: 252 seconds)
16:32:06  * caiolima__quit (Ping timeout: 252 seconds)
16:32:08  * akleinjoined
16:32:10  * mmun_changed nick to mmun
16:34:01  * caiolima__joined
16:34:42  * brianloveswordsjoined
16:49:05  * Fishrock123quit (Remote host closed the connection)
16:53:54  * Fishrock123joined
16:56:58  * Fishrock123quit (Remote host closed the connection)
17:01:31  * jeffmoquit
17:01:45  * jeffmojoined
17:11:23  <TabAtkins>dilijev: Some other working groups (CSSWG, for example) actually do take true transcriptions. Live, so things do occasionally get missed and stuff is lightly editted by the minute-taker, but still.
17:11:55  <TabAtkins>(We originally farmed out the responsibiity to WG members, but now I have someone on contract that does it for us.)
17:46:52  * gskachkov_joined
17:56:39  * gskachkov_quit (Quit: gskachkov_)
17:57:53  * Fishrock123joined
18:05:09  * not-an-aardvarkjoined
18:18:08  * gskachkov_joined
18:35:33  * bradleymeckquit (Quit: bradleymeck)
18:40:16  * bradleymeckjoined
18:46:51  * bradleymeckquit (Quit: bradleymeck)
18:48:48  * gskachkov_quit (Quit: gskachkov_)
18:54:38  * gskachkov_joined
19:10:19  * gskachkov_quit (Quit: gskachkov_)
19:17:15  * gskachkov_joined
19:21:25  <ljharb>TabAtkins: tbh if a contracted notetaker had the technical expertise to do it, and we retained the ability to edit prior to publication, that might be nice to have
19:22:52  <TabAtkins>Yeah, finding someone with enough expertise to accurately transcribe a technical meeting is the hard part.
19:24:27  * bradleymeckjoined
19:24:31  * ErikCorryquit
19:24:45  * ErikCorryjoined
19:53:29  * gskachkov_quit (Quit: gskachkov_)
19:56:48  * gskachkov_joined
20:08:53  * not-an-aardvarkquit (Quit: Connection closed for inactivity)
20:09:28  * Fishrock123quit (Remote host closed the connection)
20:14:38  * Fishrock123joined
20:37:07  * Fishrock123quit (Remote host closed the connection)
20:37:47  * Fishrock123joined
20:37:55  * Fishrock123quit (Remote host closed the connection)
20:38:37  * Fishrock123joined
20:38:43  * Fishrock123quit (Remote host closed the connection)
20:53:17  <bterlson>TabAtkins: ljharb: I am talking now w/ ECMA about whether we can pay someone to take minutes. It would be so nice.
21:23:00  <cloudshu>ooo
21:23:08  <cloudshu>like a court reporter??
21:26:27  <cloudshu>bterlson: did you ever get around to adding a flag that'll inline annex b in the output?
21:26:31  <cloudshu>bterlson: to ecmarkup, i mean
21:27:50  <bterlson>cloudshu: no
21:28:01  <cloudshu>bterlson: still planning to?
21:28:08  <cloudshu>bterlson: would be mighty nice
21:28:09  <bterlson>yes
21:28:12  <cloudshu>\o/
21:28:13  <bterlson>but I feel like
21:28:16  <bterlson>it's obviously so nice
21:28:19  <bterlson>we should just do i
21:28:20  <bterlson>it
21:28:22  <cloudshu>i mean, yes
21:28:31  <bterlson>and not make poor overworked bterlson do tooling work to give us what we should just have ;)
21:28:32  <cloudshu>but the flag is obviously less controversial
21:28:53  <bterlson>but yes
21:28:55  <bterlson>I will do the flag
21:29:15  <cloudshu>bterlson: actually, well, is inlining it itself uncontroversial?
21:29:31  <cloudshu>bterlson: that is, if we sidestep any of the normative optional, should annex b exist questions
21:29:39  <cloudshu>bterlson: and just inlined it with different styling
21:29:56  <cloudshu>bterlson: if that's uncontroversial, i can submit a PR to just do that
21:29:58  <bterlson>cloudshu: I think it was yes
21:30:02  <bterlson>cloudshu: review the notes
21:30:08  <cloudshu>bterlson: ok cool
21:30:18  * gskachkov_quit (Quit: gskachkov_)
21:30:20  <bterlson>Having it be separate gives it an air of otherness
21:31:18  <cloudshu>bterlson: well then it should be renamed Annex A for Autre
21:31:48  <bterlson>it's B for Baggage
21:32:03  <cloudshu>that's pretty good
21:32:38  <bterlson>Look at how many G's baggage has
21:32:42  <bterlson>is that even a word?
21:33:08  <cloudshu>it's probably french
21:37:38  * Fishrock123joined
22:19:18  * not-an-aardvarkjoined
22:25:53  * bradleymeckquit (Quit: bradleymeck)
22:34:56  <ljharb>bterlson: it would be super nice, provided expertise + being able to edit still
22:35:56  * bradleymeckjoined
22:42:16  * Fishrock123quit (Quit: Leaving...)
23:10:22  <dilijev>ljharb: TabAtkins: bterlson: as for recordings of meetings -- from personal experience as secretary of a board, having a phone taking a recording of the meeting takes the pressure off of me to take perfectly accurate notes during meetings. Sometimes you don't even know what is most relevant until after the conclusion of an agenda item or vote. The
23:10:23  <dilijev>understanding is that these recordings will be used only by me to improve and fill in the minutes after the meeting when I prepare them for general publication to the board and public. That is a huge help for me to actually be present for important discussions without putting my duty to take minutes strictly above my ability to give input.
23:11:04  <dilijev>ljharb: TabAtkins: bterlson: that said, i'm talking about meetings of a few hours and I can usually home in on things that need flushing out pretty easily. Very different from a multi-day (all-day) meeting.
23:46:06  <ljharb>from experience, it's really difficult to listen to an entire TC39 meeting day after the fact, even when i'm interested in the proceedings
23:46:35  <ljharb>(once i called in, and made a recording so i could go to sleep and listen to it when i woke up. it wasn't the best way to attend)
23:54:09  * cmathesonpart ("WeeChat 1.7")