| 01:20 | <ljharb> | rbuckton: re resolve, it definitely will get support for it. Just haven’t had time to finish |
| 04:17 | <ljharb> | Bakkot: is `async do { do { return } }` currently a syntax error? if so, do you know how to go about doing that? (i'm assuming some kind of "async do mode" something or other) |
| 04:17 | <Bakkot> | yes, and the way you do this is just by parsing the body of `async do` with `[~Return]` |
| 04:52 | <ljharb> | aha, thanks |
| 13:55 | <robpalme> | and the Jitsi is open for business |
| 14:06 | <michaelficarra> | I find the difference in behaviour from what the champions intended to be a compelling argument |
| 14:21 | <leobalter> | https://gist.github.com/leobalter/16364bb167633cb3cb31e0f95e160a2a |
| 14:27 | <ystartsev> | I feel like... it would be really interesting to merge test262 and engine262 -- when a spec is ready to go to stage 3, it can have a direct implementation in engine262 and have tests written with it |
| 14:27 | <ystartsev> | thinking aloud |
| 14:28 | <devsnek> | i'm a fan of reference impls |
| 14:28 | <devsnek> | we have one for wasm |
| 14:28 | <shu> | ystartsev: i'm... skeptical? |
| 14:28 | <shu> | who's reviewing the engine262 implementation? |
| 14:29 | <shu> | like is that another thing to slow velocity and increase burden on the committee delegates, who are already volunteers? |
| 14:29 | <ystartsev> | I believe at the moment it is devsnek, but this could be a shared responsibility |
| 14:29 | <shu> | i'd rather not have that kind of bus factor |
| 14:29 | <michaelficarra> | if engine262 could be generated from the spec then I would support it, otherwise it's too much of a risk for additional divergence |
| 14:29 | <shu> | +1 |
| 14:29 | <ystartsev> | we had work to generate a parser from spec text in spidermonkey |
| 14:30 | <ystartsev> | that is paused... but there is a non-zero chance that we will continue it |
| 14:30 | <devsnek> | jmdyck has a number of static analysis tools for the spec |
| 14:30 | <michaelficarra> | we can ask the INRIA people how feasible they think it'd be at the moment |
| 14:30 | <devsnek> | but they aren't public |
| 14:30 | <shu> | ystartsev: a grammar is one thing, but the algorithm steps is another |
| 14:30 | <ystartsev> | in terms of implementing semantics, that will be a bit ore difficult |
| 14:30 | <ystartsev> | *more |
| 14:31 | <ystartsev> | it would also reduce how vague/general we can be in spec text. for example how we wrote the top level await spec change |
| 14:31 | <ystartsev> | which would not be good, imho |
| 14:31 | <michaelficarra> | ystartsev: doing that is already one of my top priorities as editor |
| 14:32 | <ystartsev> | as in, make the algorithm steps more specific? |
| 14:32 | <michaelficarra> | generating an impl from the spec is effectively my end goal |
| 14:32 | <devsnek> | would be an interesting project |
| 14:32 | <ystartsev> | hm, ok |
| 14:32 | <ljharb> | that would be amazing |
| 14:32 | <ljharb> | would generating test cases from the spec be harder or easier? |
| 14:32 | <ystartsev> | so one concern I have there is that we may lose the intention of a given segment of spec in favor of an implementation |
| 14:32 | <michaelficarra> | ljharb: should be just about the same, look into abstract interpretation |
| 14:33 | <ljharb> | k |
| 14:33 | <ryzokuken> | wanted to clarify that by "discussion" I mean the one Shu mentioned and that we should kick off an async thread somewhere that talks about improvements to test262 |
| 14:33 | <ystartsev> | maybe this can be addressed through editorial notes though |
| 14:33 | <michaelficarra> | ystartsev: agreed, we cannot lose human-readability |
| 14:34 | <ystartsev> | I think this is one issue of the top level await specification right now, it is too step oriented without enough text explaining the intention of these rather long graph algorithmsm |
| 14:34 | <ystartsev> | we _do_ have an entire non-normative example segment of the spec |
| 14:34 | <ryzokuken> | ljharb: I was thinking down the spec -> tests line as well |
| 14:34 | <ryzokuken> | maybe not the spec itself |
| 14:34 | <ystartsev> | but it would be better to have something in context with the algorithm |
| 14:34 | <haxjs> | it seems the audio quality is not very good... |
| 14:35 | <ryzokuken> | but there could certainly be a test262 generator of sorts |
| 14:35 | <ystartsev> | haxjs: maybe try reloading? |
| 14:35 | <ljharb> | ystartsev: +1 that the TLA spec is almost entirely unapproachable, but i have no idea how to fix that |
| 14:35 | <shu> | ystartsev: yeah i don't think test262 is even going to solve that |
| 14:35 | <ystartsev> | ljharb: I want to go through it at some point and bring some intentional text around it, either through notes or otherwise |
| 14:35 | <shu> | ystartsev: it's very telling that the discrepancies came up during fuzzing, right? |
| 14:35 | <haxjs> | ystartsev: serveral chinese delegates report same audio issue... |
| 14:35 | <ystartsev> | shu: yes |
| 14:36 | <leobalter> | shu: your suggestion requires browsers to commit with a interoperable runner, I believe so |
| 14:36 | <leobalter> | it's a chicken and egg problem |
| 14:36 | <devsnek> | i'm curioust |
| 14:36 | <ljharb> | haxjs: hm, maybe an audio latency thing |
| 14:36 | <devsnek> | curious* |
| 14:36 | <ystartsev> | i like the idea that shu brought up |
| 14:36 | <shu> | leobalter: no it does not |
| 14:36 | <devsnek> | there's currently an "implementation contributed" dir in test262 |
| 14:36 | <devsnek> | what's the status of that |
| 14:36 | <ystartsev> | we have a test262->sm converter but not the other way around |
| 14:36 | <shu> | devsnek: no one uses it |
| 14:36 | <devsnek> | ah ok |
| 14:36 | <shu> | leobalter: everyone already has a runner shim that works for them |
| 14:37 | <leobalter> | shu: everyone's runner has different perks that makes it so hard to change things in test262 |
| 14:37 | <shu> | leobalter: i disagree, they don't really matter for the tests for the vast majority of tests? |
| 14:37 | <ystartsev> | haxjs: is it still bad after refreshing? |
| 14:38 | <shu> | once in a while something comes along, like weakrefs, that require new shim things to be done |
| 14:38 | <ystartsev> | cc robpalme we have some issues with audio for the chinese delegates |
| 14:38 | <robpalme> | right now? |
| 14:38 | <leobalter> | the Test262 python runner used in v8 is the hardest to change any frontmatter thing today. Relaxing it would be a great start to change interoperability. |
| 14:38 | <leobalter> | shu ^^ |
| 14:39 | <shu> | leobalter: could you say more? |
| 14:39 | <shu> | leobalter: you're talking about the copyright checker? |
| 14:40 | <leobalter> | the copyright is one thing, the frontmatter checker is very strict |
| 14:40 | <michaelficarra> | leobalter: isn't it open source? It shouldn't be that hard to suggest a change. |
| 14:40 | <shu> | leobalter: oh really? let's get rid of it then :) |
| 14:40 | <shu> | i am not exaggerating when i say i derive basically no value from that frontmatter |
| 14:40 | <leobalter> | michaelficarra: there is a separate python runner that is a clone-like of V8's runner |
| 14:41 | <robpalme> | how is the audio now? |
| 14:41 | <devsnek> | frontmatter = the comment at the top with the spec steps and stuff? |
| 14:41 | <leobalter> | IMO, verifying frontmatter should be a task from test262's lint |
| 14:41 | <haxjs> | ystartsev: I've reloaded but still not good, not sure about others. |
| 14:42 | <leobalter> | we need to make sure we have a frontmatter for things that used in tests (e.g. flags) |
| 14:42 | <ystartsev> | hm ... we also had the audio muted for all others |
| 14:42 | <leobalter> | devsnek: yes |
| 14:42 | <robpalme> | haxjs: can you try a different browser? or use one of the native apps? |
| 14:43 | <haxjs> | Oh I don't know there is native app! Let me try :) |
| 14:51 | <littledan> | wsdferdksl: Can you back up that assertion about NVC? It's possible that I'm missing background here. |
| 14:52 | <ljharb> | flags, a description, and a link to the relevant part of the spec, seem like the only parts of the frontmatter that would be useful, but i certainly don't know everyone's use case |
| 14:53 | <wsdferdksl> | littledan: From cnvc.org: "I think it is important that people see that spirituality is at the base of Nonviolent Communication, and that they learn the mechanics of the process with that in mind. It's really a spiritual practice that I am trying to show as a way of life. Even though we don't mention this, people get seduced by the practice. Even |
| 14:53 | <wsdferdksl> | if they practice this as a mechanical technique, they start to experience things between themselves and other people they weren't able to experience before. So eventually they come to the spirituality of the process. They begin to see that it's more than a communication process and realize it's really an attempt to manifest a certain spirituality." |
| 14:53 | <wsdferdksl> | That's a quote from Rosenberg. |
| 14:54 | <ystartsev> | rosenberg is seen as a complicated figure in terms of communication theory, he is coming from a very particular history, very much based in the usa |
| 14:55 | <ystartsev> | I think that, yes, for him it was rooted in spirituality specifically from the 60s and 70s when it was developed |
| 14:56 | <ystartsev> | however, it isn't exclusively a spiritual practice, and there are no spiritual aspects to the training itself. It focuses primarily on contextualizing a situation between two people during communication |
| 14:56 | <ystartsev> | indeed, on cnvc, they say "NVC *can* be seen as both a spiritual practice that helps us see our common humanity, (...), *and* a concrete set of skills" |
| 14:57 | <michaelficarra> | it sounds like you're painting it as historical, but it's taken from the website linked to by the presented slides |
| 14:57 | <ystartsev> | We can look at alternative types of communication strategies, however NVC is one of the older ones and more tested. It isn't perfect so if you have alternatives to suggest, I think people would be open to that |
| 14:57 | <ystartsev> | michaelficarra: yes, it is taken from the website, as is the quote i added above |
| 14:58 | <ystartsev> | eating can also be a spiritual activity, yet we have dinners as a group. |
| 14:59 | <ljharb> | one of the criticisms of modern yoga practice is that it *isn't* spiritual as yoga was originally created to be; is it necessarily important what the roots of the thing are, if we're not following those roots? |
| 15:03 | <littledan> | Donald Knuth relates algorithms to Christianity, but we still benefit from his work |
| 15:04 | <shu> | whatever communication training is decided upon, the ultimate question i imagine is what % of delegates will actually be open to it |
| 15:04 | <rbuckton> | Is jitsi working? I've tried on two different machines and hear no sound. |
| 15:05 | <michaelficarra> | I'm definitely still open to it, but the mentions of spirituality on the website do make me uncomfortable |
| 15:05 | <michaelficarra> | rbuckton: yes |
| 15:05 | <ystartsev> | What i like about nvc is it might help us drill down to the invariants and values that we want to maintain about the language. I am actually not 100% behind nvc training, but individual trainers interpret the method differently |
| 15:08 | <leobalter> | rbuckton: I'm using the phone app for audio |
| 15:08 | <rbuckton> | So far I've tried two different PCs using the electron app and the web and I'm not hearing audio. :/ |
| 15:09 | <michaelficarra> | rbuckton: did you try turning them off and then on again? |
| 15:09 | <rbuckton> | michaelficarra: yep, and its plugged in |
| 15:16 | <rbuckton> | I think I'm having local internet issues :/ |
| 15:17 | <rbuckton> | unless tcq is also down. Guess I'll blame comcast. |
| 15:28 | <rbuckton> | comcast was the issue. |
| 15:30 | <littledan> | ljharb: Are you saying you'd block this from joining the proposals repo until it has a problem statement? |
| 15:30 | <ljharb> | no, i'd have used the word "block" |
| 15:32 | <littledan> | well, I was just confused by what you meant when you talked about how the proposals repo ties into this |
| 15:32 | <ljharb> | but we've discussed this in plenary before. bringing two proposals that have solutions for stage 1 doesn't match that. |
| 15:32 | <ljharb> | that was my primary motivation for asking |
| 15:32 | <ljharb> | my lack of understanding of this presentation isn't as important to me since i don't often pay attention to typed array stuff |
| 15:33 | <ljharb> | but it is a frequent request of proposals seeking stage 1, by many delegates, so i'm surprised by your pushback |
| 15:33 | <littledan> | I think I phrased my pushback poorly |
| 15:34 | <littledan> | the whole presentation was about the motivation, which satisfies the Stage 1 requirements |
| 15:34 | <ljharb> | sure. but the slide asking for stage 1, and the proposal repo name and readme title, are phrased in terms of 2 competing solutions |
| 15:34 | <ljharb> | i'm happy it was apparently clear to everyone else, but it wasn't to me. |
| 15:35 | <michaelficarra> | this is a frequently recurring problem, especially among new delegates that are less familiar with our process |
| 15:35 | <michaelficarra> | maybe we need to communicate our process more effectively |
| 15:37 | <michaelficarra> | like could the proposal template repo encourage starting with a problem statement instead of jumping into a solution, spec text, etc? |
| 15:37 | <littledan> | I like that suggestion |
| 15:38 | <Bakkot> | devsnek jmdyck's static analysis tools are in fact public: https://github.com/jmdyck/ecmaspeak-py |
| 15:39 | <devsnek> | oh ni |
| 15:39 | <devsnek> | nice |
| 15:39 | <Bakkot> | I had a working knowledge of it at some point though I haven't done anything recently |
| 15:40 | <ljharb> | michaelficarra: that sounds like a great improvement |
| 16:00 | <littledan> | Hmm, from the 402 notes, I do agree with Myles that we should collaborate with Open UI, even if we do this in 402 |
| 16:02 | <littledan> | I'm having trouble finding the part of the notes where anyone from Apple said that this proposed API didn't result in a fingerprinting vector; maybe someone can point to the exact location? |
| 17:01 | <shu> | sorry i joined late due to another meeting |
| 17:01 | <shu> | what were the options? |
| 17:01 | <akirose> | we'll decide after this |
| 17:02 | <shu> | but what were the options that yulia presented? |
| 17:02 | <Bakkot> | make this one 30 minutes longer or get up tomorrow morning or something else |
| 17:02 | <ystartsev> | shu: options are open right now |
| 17:02 | <akirose> | the options are going 30 min late today, or meeting for just 30 min tomorrow |
| 17:02 | <shu> | ah, 30 mins today preferred then |
| 17:02 | <ystartsev> | 30 min late today, 30 min tomorrow starting at 10 am, or after lunch 30 minutes |
| 17:02 | <Bakkot> | I would strongly prefer not to get up in the morning again tomorrow |
| 17:02 | <shu> | both 30 mins w/o break or 30 minutes after lunch wfm |
| 17:02 | <Bakkot> | am ok with other options but prefer running over today |
| 17:03 | <shu> | same |
| 17:05 | <ryzokuken> | I did see the slides previously too... |
| 17:05 | <ryzokuken> | but yeah |
| 17:05 | <ljharb> | same |
| 17:05 | <ystartsev> | do we have note takers? |
| 17:05 | <rpamely> | I think there's a bug on the mobile app where it only shows slides if the share started while you were connected. Not sure if people not seeing it were on mobile. |
| 17:06 | <msaboff> | Running 30 minutes over today is preferred. |
| 17:07 | <ryzokuken> | +1 for running over but if we have other APAC delegates online, maybe we should confirm from them? |
| 17:07 | <ystartsev> | I am checking and will explicitly check before we make a decision |
| 17:07 | <ryzokuken> | awesome thanks |
| 17:08 | <ryzokuken> | it's the jitsi join/leave sound, can be disabled IIRC |
| 17:13 | <akirose> | hey friends when Leo is done presenting slides, we'll need another notetaker (or two) |
| 17:13 | <akirose> | this is your heads up |
| 17:14 | <rricard> | I unfortunately can't continue further today |
| 17:14 | <akirose> | rricard: you have done so much, thank you ♥️ |
| 17:15 | <littledan> | apologies for the distraction before; ryzokuken found the place where Myles Maxfield of WebKit explicitly supported Mozilla's privacy analysis. Still, Apple had other ideas about how the API should be shaped, and we should probably discuss these further (in TG2). |
| 17:18 | <ystartsev> | should i call for notes? |
| 17:32 | <littledan> | let me know if anyone wants a link to the April 2021 TG2 meeting (which wasn't linked from the calendar like normal). It contains answers to a lot of the questions that I've (inappropriately) asked during this meeting about Intl proposals. Again, my apologies for the redundancy/confusion of my comments. |
| 17:42 | <devsnek> | i dislike that virtualization apis would be subject to seemingly random limitations |
| 17:42 | <devsnek> | i can always make a version of evaluate that doesn't allow objects, but i can't make something the other way around |
| 17:51 | <littledan> | I think Realms are quite a poor replacement for getOriginals tbh. You want originals from the same Realm! |
| 17:52 | <jridgewell> | Yah, I think all of the internal slots won't be accessible with the user-realm's function references |
| 17:52 | <Bakkot> | strong disagree; getOriginals was very bad and I am extremely glad it did not go anywhere |
| 17:53 | <jridgewell> | Eg, `userRealm.WeakMap.prototype.get.call(incubatorWeakMap)` is gonna throw |
| 17:53 | <littledan> | well, there are legitimate concerns about getOriginals, but I don't think it's a good pattern to make a new Realm to replace it. |
| 17:53 | <ljharb> | getOriginals as proposed was indeed very bad |
| 17:54 | <ljharb> | and i do not ever want originals from the same Realm, in fact. |
| 17:54 | <ljharb> | jridgewell: no it won't? |
| 17:54 | <ljharb> | jridgewell: internal slots are cross-realm |
| 17:55 | <shu> | littledan: strong +1 |
| 18:00 | <devsnek> | i wish the realms proposal would be more 1:1 |
| 18:00 | <devsnek> | let me fully virtualize js environments :( |
| 18:01 | <ljharb> | that's basically what it was, for years |
| 18:01 | <jridgewell> | ljharb: Hmm, that's surprising. |
| 18:02 | <jridgewell> | c |
| 18:03 | <ljharb> | jridgewell: it's what Array.isArray is for, because `instanceof Array` doesn't work cross-realm, but slot-checking does |
| 18:15 | <littledan> | jridgewell: can't you use CSP to ban that Blob API too? |
| 18:15 | <littledan> | I mean, blob URLs |
| 18:16 | <Bakkot> | no one does, though |
| 18:16 | <littledan> | :( |
| 18:16 | <Bakkot> | lot of people disable eval, and the csp spec encourages that, but no one disables blobs |
| 18:16 | <Bakkot> | they don't have nearly the same issues |
| 18:16 | <Bakkot> | (since they don't evaluate in the context of the page you're on) |
| 18:17 | <littledan> | well, this proposal doesn't require new CSP hooks; it should fit into the existing ones in the same way |
| 18:17 | <rickwaldron> | @jridgewell here's an early discussion: https://github.com/tc39/notes/blob/8711614630f631cb51dfb803caa087bedfc051a3/meetings/2013-11/nov-21.md |
| 18:17 | <rickwaldron> | I believe that was the first time Dave presented this: https://gist.github.com/dherman/7568885 |
| 18:18 | <Bakkot> | oh I misunderstood, the way amp is using it would be same-realm I think |
| 18:19 | <ryzokuken> | ystartsev: I'll go |
| 18:19 | <jridgewell> | We purposefully don't ban `blob:` in CSP, to get around the fact that we ban `unsafe-eval` |
| 18:20 | <Bakkot> | jridgewell lol it did not even occur to me that you would be in a position to control that yourselves |
| 18:20 | <Bakkot> | I am so used to the CSP being owned by a different team |
| 18:20 | <jridgewell> | This is the reason for the AMP Cache |
| 18:20 | <jridgewell> | (Partly) |
| 18:20 | <Bakkot> | (as it is in, afaict, ~every company other than google) |
| 18:21 | <ystartsev> | ryzokuken: thank you |
| 18:21 | <Bakkot> | (which, it is definitely designed for the google approach) |
| 18:53 | <brad4d> | I'll volunteer to review Symbols as WeakMap keys |
| 18:53 | <devsnek> | 🎉 |
| 18:54 | <rickbutton> | i got kicked out of #tc39-delegates, weird |
| 18:54 | <ljharb> | fired, obvs |
| 18:54 | <rickbutton> | oh no |
| 19:11 | <Bakkot> | ystartsev I can do notes |
| 19:13 | <ystartsev> | thanks Bakkot |
| 19:20 | <leobalter> | are we discussing test262 today? I logged off to decompress, I'd be happy to rejoin in this case |
| 19:21 | <littledan> | ? we discussed test262 in the morning |
| 19:21 | <leobalter> | it appears in the overflow list |
| 19:21 | <ljharb> | i think we're wrapping up overall |
| 19:21 | <leobalter> | I'm just checking |
| 19:21 | <ljharb> | if you want to talk about it for the last 9 minutes please speak up |
| 19:21 | <leobalter> | I don't |
| 19:21 | <ljharb> | kk |
| 19:21 | <leobalter> | I'm exhausted |
| 19:22 | <leobalter> | but glad, for the records |
| 19:22 | <littledan> | it doesn't appear in the overflow list for me |
| 19:22 | <robpalme> | we're not going back to test262 - that record of the queue is just for gus |
| 19:22 | <leobalter> | thanks for the clarification, robpalme |
| 19:23 | <littledan> | finally we'll adopt Ecma's practices and schedule our meetings for longer than they actually are so that we can justify travel, more time, etc |
| 19:24 | <akirose> | lol/wince |
| 19:24 | <wsdferdksl> | The General Assembly meetings are scheduled for two days but always manage to finish in one. |
| 19:27 | <leobalter> | I would not oppose to schedule a fifth day for socializing. It still shows in the calendar for traveling expenses but it could remain optional for individuals |
| 19:27 | <leobalter> | no need to lie, we could always schedule social activities |
| 19:30 | <akirose> | gap day like jsconfus |
| 19:31 | <akirose> | "everyone take a breath" |
| 19:31 | <ryzokuken> | anyone coming to hubs? |
| 19:32 | <ljharb> | free time sounds nice *cries in calendar* |
| 19:36 | <ryzokuken> | cries in .ics :P |
| 19:43 | <rickbutton> | +1 for a gap day |
| 19:43 | <rickbutton> | especially if we go back to hawaii ;) |
| 19:45 | <Bakkot> | I took a week off work after that meeting |
| 19:45 | <Bakkot> | was the last time I was on a plane |
| 19:45 | <Bakkot> | in retrospect, great timing |
| 19:45 | <ljharb> | sames |
| 20:41 | <shu> | yes that was also the last time i was on a plane |
| 23:10 | <TabAtkins> | I was already quarantining by then :( |
| 23:10 | <TabAtkins> | But my last time on a plane was in Spain the month prior, so hey, still great. |
| 23:14 | <rkirsling> | I deeply regret not going to hawaii |