2022-11-02 [15:56:33.0083] PSA: we just landed a commit on ecma262 which tweaks the formatting, which is the sort of thing which causes a lot of merge conflicts. if you have an in-flight PR and you're rebasing on main, you can avoid manually applying the formatting by using the following series of commands: ``` git rebase -i 0090daf^ # 0090daf is the commit where the formatter is applied git rebase -X theirs 0090daf --exec "npm i && npm run format && git commit --amend -a --no-edit" git rebase -i main ``` this splits up the rebase into 1) everything before the formatter was applied, 2) the formatter, which will be done for you, 3) everything after the formatter. any conflicts in parts 1 and 3 you will need to handle manually, just like you would in the absence of the formatter (or feel free to ask the editors for help) 2022-11-07 [10:21:20.0830] yulia: are we unshipping `group`? [10:21:26.0878] * yulia: are we unshipping `group`? [10:49:02.0930] oof, i hope not - let's at least see if we can fix lastpass first [10:50:46.0642] hm, even if we can i'm still leaning towards unshipping in chrome [10:51:07.0420] because i assume it'll take a while for lastpass's fix to deploy to a large enough % of its customers [10:53:40.0259] i'd still want to know what code was breaking on it [10:54:32.0347] +1, the bugzilla comment reads to me as a suspicion, i want to hear it confirmed first [10:54:53.0377] (i mean it sounds likely, but still) [10:57:37.0180] There is a risk that this may be very hard to catch in codebases, but maybe we just got unlucky with this [10:58:10.0911] Since chrome is about to ship I didn't think it would be wise to wait on notifying you [10:58:53.0815] indeed, we have a window of a week until i need to request stable respins, which is not good [10:59:13.0750] so much appreciated there [11:00:29.0058] maybe this is the chance to get `groop` after all [12:12:05.0246] shu: ljharb it... isn't looking good. Two more regressions: https://github.com/tc39/proposal-array-grouping/issues/44 [12:19:51.0959] 😭 [12:20:58.0470] Is this the same as that Lego site’s using an array as a hash map? [12:59:00.0192] the cloud.ibm.com one appears to be, though it's different code [13:41:05.0245] yulia: thanks! [13:41:13.0432] alas, i am unshipping this [14:27:57.0824] sigh [14:28:19.0179] that was the only good other name [14:30:35.0401] there's always `smoosh` [14:30:49.0364] what about `bucket` [14:30:56.0867] or the more provocative `buckit` [14:31:22.0927] `smooshIntoBuckets` [14:31:25.0190] buck it, we're doing 5 groups [14:33:20.0146] I has a bucket. [14:33:57.0715] the array-as-hash-table use case makes name picking pretty unpredictable [14:34:01.0969] who knows what keys they're using [15:56:51.0447] Let's introduce a new meta-object protocol called `getAndApply`, and update `Array.p.group` to: 1. Returned undefined for `get` where the receiver is an array instance 2. Return the function for `get` where the receiver is `Array.prototype` 3. Invokes the method when doing `[].group()` (`getAndApply`) [15:57:17.0188] Then any new array methods can just be hidden from these terrible sites. 2022-11-08 [16:04:49.0314] or.... we just choose a different name [16:05:09.0039] and eventually stop trying to put new things on `Array.prototype` [16:05:27.0416] hopefully one day we can have built-in modules or something and then build out a nice stdlib [16:05:35.0171] yeah i do not think we should introduce a new MOP protocol for this [16:06:15.0885] Symbol.facetious [16:06:23.0783] quite right, we should use the existing `document.all` protocol [16:06:59.0102] I was waiting for someone to bring up `IsHTMLDDA` [16:27:02.0402] We pre-scan the source text to see if the word “group” appears anywhere, and if so, we do not install the method. Group is only accessible through computed property access were the key is not literally the string “group”. [18:16:00.0900] > <@shuyuguo:matrix.org> who knows what keys they're using maybe the engine need to record names that added to the array happens in Web [18:18:17.0779] > <@jridgewell:matrix.org> We pre-scan the source text to see if the word “group” appears anywhere, and if so, we do not install the method. Group is only accessible through computed property access were the key is not literally the string “group”. and developers need to use it by `array["gro" + "up"]()` right [18:30:07.0994] > <@jackworks:matrix.org> maybe the engine need to record names that added to the array happens in Web that sounds way too expensive to be feasible [18:30:12.0909] also privacy and fingerprinting concerns? [00:58:26.0471] > <@shuyuguo:matrix.org> that sounds way too expensive to be feasible I thought the engine was already doing that today, e.g. metrics of Web API usage [08:07:13.0378] those are known counters, they aren't tracking new strings [12:32:43.0376] yulia: is Nightly shipping change-array-by-copy? [12:33:23.0528] `toSorted`, `toSpliced`, and `toReversed` seem unlikely to be used as keys [12:33:29.0034] but i am not so sure about `with` [12:56:13.0062] `with` seems a lot less likely than `group` [13:00:48.0369] > <@shuyuguo:matrix.org> yulia: is Nightly shipping change-array-by-copy? only behind a build time flag [13:32:32.0830] okay, thanks [13:32:50.0427] i tried looking for the ibm issue and the other issue on the chromium issue tracker and didn't see anything [13:34:10.0099] canary population is pretty small, but i wonder if there're biases in play with people not filing issues, which would be unfortunate [13:34:49.0623] > <@bakkot:matrix.org> `with` seems a lot less likely than `group` i agree, guess we'll see 2022-11-09 [06:22:58.0899] yulia | 🤕 hopefully back in the afternoon: Do you think either you or someone from your organization would have some time in the next few weeks before plenary to discuss Mozilla's position on the Explicit Resource Management proposal? [06:23:57.0387] Sure, we discussed before the last presentation. dminor and I can set something up [06:24:43.0236] before next plenary or? (we will also be in a coruna) [06:24:50.0387] That's fine. As far as scheduling, I'm on the US east coast now. [06:25:17.0165] Before, if possible. I'd like to be able to advance if that's feasible. [06:27:07.0255] ok, i'll set something up 2022-11-10 [17:10:38.0989] anyone interested in running an incubator call for BigInt math (round 2) before the next plenary? 2022-11-14 [14:47:04.0181] ljharb: the tc39 calendar has the plenary as 7 hours long, which is presumably a mistake 2022-11-15 [16:04:21.0486] fixed i think - since it’s in person tho isn’t that right? 10-5? [16:04:54.0262] I'm just going off of the agenda [16:05:59.0758] ah ok they’re 10-4 [16:10:58.0701] looking forward to falling asleep mid-sentence yall [11:02:21.0081] Apologies to anyone trying to join the R&T monthly meeting. The zoom link appears to have stopped working. new link in the google doc, linked from the tc39 event 2022-11-16 [04:19:58.0054] 2023 Plenary Schedule is now posted on [the Reflector](https://github.com/tc39/Reflector/issues/453). There's just one date missing that will be confirmed very soon. 2022-11-17 [17:53:12.0244] ljharb: you've been vocal in the past about coercing arguments in the order in which they appear. is there code that you've seen that actually depends on that ordering? [17:53:56.0053] i'm thinking of proposing a breaking change to the order of coercion for Atomics methods like `Atomics.store(ta, idx, val)` to coerce idx, val, *then* validate the TA [17:54:21.0473] with resizable buffers, having to recheck detached/out-of-boundness and reloading the length after each argument coercion is really unfortunate [17:55:18.0726] it's kind of a bugfarm (sometimes security bugs) [18:31:23.0879] no code i think the committee would respect, I’d guess. why is it a bug farm? [18:31:39.0838] I’m confused about the problem it’d be solving [18:35:48.0758] Agree with shu. Don't think coerce order will break anything [18:45:03.0984] it's a bug farm because it's easy to forget that coercing `idx` and `val` can detach `ta` [18:45:15.0655] (and with resizable buffers, can resize `ta`) [18:46:03.0551] TA detachment and resizing is a qualitatively worse kind of user code than other arbitrary user code, because code that forget to recheck for detachedness after each argument coercion can end up reading out-of-bounds into the buffer, depending on the implementation [18:46:12.0006] * TA detachment and resizing is a qualitatively worse kind of user code than other arbitrary user code, because code that forget to recheck for detachedness after each argument coercion can end up reading out-of-bounds into the buffer, depending on the implementation [18:46:20.0487] "easy to forget" do you mean when v8 is optimizing code? [18:46:41.0711] that, and i also mean engineers or spec authors actually forgetting even when implementing the slow path [18:46:55.0241] i personally missed a bunch of these in the resizable buffers spec draft [18:48:26.0482] TA and Atomics methods depend on the TA being in a valid state, life will just be a lot easier if we can rework these methods to 1) validate once instead of multiple times and 2) after validation, check that no user code can ever be called again until method exit [18:48:59.0753] * TA and Atomics methods depend on the TA being in a valid state, life will just be a lot easier if we can rework these methods to 1) validate once instead of multiple times and 2) after validation, check that no user code can ever be called again until method exit [18:50:00.0631] > <@ljharb:matrix.org> I’m confused about the problem it’d be solving what i'm solving for is "increasing likelihood of correct implementation, where incorrect implementation is often a security bug and not just misbehavior" [18:51:52.0837] to be clear, user code re-entrancy is a general problem. i'm singling out TAs because TA bugs result in more, and more serious, security bugs [18:54:07.0978] So the ideal order is: idx check, val check, (no user code now), TA check, Atomics.store, (user code again)? [18:54:37.0225] it's a little trickier than that unfortunately, because `val` coercion depends on looking at whether `ta` is a BigInt TA or a non-BigInt TA [18:55:02.0921] so i think the ideal order is: coerce *but not validate* left-to-right, then validate left-to-right [18:55:54.0505] lgtm [18:55:56.0734] so the order is: 1) ensure `ta` has the right internal slots, 2) coerce `idx`, 3) coerce `val`, (no user code after this point) 4) validate `ta`, 5) validate `idx` (if needed for atomic access) [18:56:08.0694] * so the order is: 1) ensure `ta` has the right internal slots, 2) coerce `idx`, 3) coerce `val`, (no user code after this point) 4) validate `ta`, 5) validate `idx\` (if needed for atomic access) [18:56:17.0296] * so the order is: 1) ensure `ta` has the right internal slots, 2) coerce `idx`, 3) coerce `val`, (no user code after this point) 4) validate `ta`, 5) validate `idx` (if needed for atomic access) [19:45:31.0834] in general, does anything validate besides TAs and Proxy? [19:45:40.0016] * in general, does anything validate besides TAs and Proxy? [22:50:18.0463] ljharb: I think that depends upon your definition of "validate". For example, there are six opportunities for [String.prototype.split](https://tc39.es/ecma262/multipage/text-processing.html#sec-string.prototype.split) to throw an exception (and also observably checks arguments out-of-order)—does that count? [06:52:58.0584] im trying to understand she’s mental model here; it’s probably fine if TA/Proxy is “special” but i still don’t understand why it’s not easier to implement correctly if you fully process args left to right [07:02:48.0197] because processing later args can invalidate previous validations [07:02:55.0893] * because processing later args can invalidate previous validations [07:30:39.0760] ljharb: yeah what Ashley Claymore said. fully left-to-right means you validate `ta`, then validate `idx`, then validate `val`, then need to validate `ta` *again* [07:31:52.0767] ahhhhh ok [07:32:06.0151] cause of, you know, `valueOf` and whatnot [07:32:11.0621] thanks, that makes sense now, let me think about it [07:33:17.0256] actually in the resizable case it's even dumber [07:33:36.0960] it's validate `ta`, validate `idx`, validate `val`, re-validate `ta`, then also re-validate `idx` because `ta` might not be out of bounds but just shrunk a little bit [07:34:38.0561] `while (somethingCouldChange) validateArgs(...args)` [08:32:27.0243] rbuckton: just checking, did you see my invite for today? [08:33:45.0873] I did not, no. [11:34:21.0193] shu: so, generally i think of "processing" arguments left to right - not necessarily validating them. meaning, i'd maybe expect something like: 1. ensure `ta` is a Typed Array 2. coerce `idx` to a number 3. coerce `val` to whatever it needs to be, if any 4. throw if TA is out of bounds or detached or whatever 5. if needed, validate `val` thoughts on that? [11:53:00.0201] yes that seems to match what i said above about an "ideal order" [11:53:11.0321] coerce left-to-right, then validate left-to-right [11:53:18.0486] where "coerce" is the phase that can call user code [11:53:33.0964] right now we don't have such a distinction, and we do both at the same time [11:58:21.0079] right. my earlier question is because i suspect most things don't actually need to "validate" (or at least, don't need to validate something that could be changed by user code post-coercion), so it doesn't come up often [12:00:51.0057] yeah [12:01:16.0203] so it sounds like you're amenable to having distinct phases like that? [12:01:20.0671] (if web compatible, of course) [12:03:12.0560] tentatively yes, i'd just want to make sure there's no weird cases that would arise from applying that rule universally (which might be fine ofc, but it's hard to know in advance) [12:03:27.0957] * tentatively yes, i'd just want to make sure there's no weird cases that would arise from applying that rule universally (which might be fine ofc, but it's hard to know in advance) [12:11:35.0350] i wasn't planning on applying it universally [12:12:03.0409] though i suspect you are right that basically only TAs (and DataViews) and proxies even have a sense of "valid state" at all [12:12:23.0110] but i only really care about TAs and DataViews 2022-11-18 [10:35:02.0136] I was thinking of changing my attendance to the upcoming plenary from remote to in-person. Is it too late to make such a change? Is there still any space at the recommended hotels? And would there be office space on Friday ? [10:40:18.0412] There is still space in Hotel Avenida, which is one of the recommended hotels. For the other questions, ryzokuken [11:39:10.0147] > <@mhofman:matrix.org> I was thinking of changing my attendance to the upcoming plenary from remote to in-person. Is it too late to make such a change? Is there still any space at the recommended hotels? And would there be office space on Friday ? I believe we have space. I'll confirm from my colleagues and ping you. [11:57:50.0648] Yes, please come in person! There are always no-shows so it would be great to see you Mathieu Hofman [13:34:18.0166] Ok great I'll fill up the in person form. [13:37:20.0158] * Ok great I'll fill up the in person form. Should I book the hotel directly or does Igalia facilitates anything? [13:39:50.0117] ryzokuken: ^ [13:45:09.0484] > <@mhofman:matrix.org> Ok great I'll fill up the in person form. Should I book the hotel directly or does Igalia facilitates anything? Yeah, we can take care of it for you. I'll reach out via DM [13:49:08.0620] @ryzokuken do we have capacity for others who still wish to sign up? [13:50:00.0685] Yeah, I'm discussing with Mathieu Hofman right now. [13:50:18.0432] ryzokuken: I think the question as about further additional people beyond Mathieu [13:50:30.0029] Oh, well, I'll have to ask. [13:51:45.0218] Rob Palmer: i think it should be doable but I'll confirm with my colleagues CET morning. [13:59:39.0508] Thank you. I hope we can maximize the number of folk who get to try out Spanish food. [14:00:00.0757] > <@robpalme:matrix.org> Thank you. I hope we can maximize the number of folk who get to try out Spanish food. As long as they can keep up with Spanish meal timings 🙈 2022-11-19 [14:59:52.0287] does anyone know how i could detach an arraybuffer in an env without postMessage or structuredClone? (non-latest node, essentially) [15:28:35.0002] Maybe via webassembly APIs? Or is that also too new? https://developer.mozilla.org/en-US/docs/WebAssembly/JavaScript_interface/Memory/grow 2022-11-20 [17:48:03.0099] ljharb: node has supported postMessage for a long time; how old a node are you talking? [17:54:06.0956] (there isn't anything in the language itself, so it'll depend on what exactly your host provides) [20:38:54.0241] um, even node 19 doesn't have postMessage? [20:39:46.0947] * um, even node 19 doesn't have postMessage? [20:40:14.0621] i mean, atm i have no solution in node at all, so wasm would be something [20:40:59.0689] node 0.5 has ArrayBuffer at least, so, as old as possible :-) [00:23:58.0059] ljharb: ah, sorry, it does but it's hard to get to [00:24:02.0586] `(new (require('worker_threads').MessageChannel)).port1.postMessage(null, [y.buffer])` [00:26:35.0286] works in 12 but not 10, but there might be a different way of getting at MessageChannel on older versions [00:30:56.0060] hm, no, nothing else I can see from a quick glance at https://nodejs.org/docs/latest-v10.x/api/all.html [00:31:03.0376] so might not be something you can do in nodes older than 12 [00:31:15.0468] (or 10 with --experimental-workers) [09:30:51.0371] thanks tho, that still helps a lot [09:31:00.0972] otherwise i was gonna have to write wasm 2022-11-22 [16:02:37.0888] > <@ljharb:matrix.org> otherwise i was gonna have to write wasm you mean otherwise you were gonna *get* to write wasm [17:54:18.0889] another thing that got reported to every browser individually: https://bugzilla.mozilla.org/show_bug.cgi?id=1800062 https://bugs.chromium.org/p/v8/issues/detail?id=13469 [17:54:59.0320] those should be closable; there might be a web reality bug there but I'm not even certain of that because engine262 matches JSC/V8/SM [18:05:06.0381] rkirsling: that's https://github.com/tc39/ecma262/issues/2659 [18:10:10.0478] nice [18:11:26.0381] thanks 2022-11-23 [11:26:29.0223] FYI: I've had to delete another abusive comment from Aaron Adams (@aadamsx on GitHub) in the pipeline repo. I'm not specifically seeking any action, because I don't want to bother, just noting it for posterity if they decide to become active again and I need to ask for a ban. [15:32:33.0283] delete, not just hide? [15:32:39.0221] we generally don't delete things 2022-11-26 [04:06:09.0636] seems like it was really good that I presented at JSConf JP; there were a number of people afterwards who were just really excited to get to talk to somebody in TC39 and ask all sorts of questions [04:06:40.0241] hopefully the Japanese JS community will have more chances for that in future :) [04:26:06.0191] There is at least one trip planned for next year. September 26-28 is the Tokyo plenary. So far we do not have a community event planned. If you have contacts, that would be helpful. 2022-11-27 [18:04:27.0552] indeed! Yosuke Furukawa is the main organizer for JSConf JP so I think he may be able to help us (IIRC littledan was even talking with him pre-pandemic about community outreach) 2022-11-28 [00:50:55.0757] This week's plenary meeting will be very busy. It is also on CEST (Madrid time) which is super early for US timezone folk. We have more content than time available. Please consider if you think any agenda items you own could use less time, or could be deprioritized, and then PR the agenda accordingly. [06:24:43.0626] FWIW, I _have_ thought about it. (I'm still basically a newbie here.) I had requested 30 minutes: 10 for my talk and 20 for discussion, as I hope for advancing my Mass Proxy Revocation proposal to stage 1. I don't have a feel for how these votes go. Even if I gave up half the discussion time, that really wouldn't buy very much. [06:26:32.0256] * FWIW, I _have_ thought about it. (I'm still basically a newbie here.) I had requested 30 minutes: 10 for my talk and 20 for discussion, as I hope for advancing my Mass Proxy Revocation proposal to stage 1. I don't have a feel for how these votes go. Even if I gave up half the discussion time, that really wouldn't buy very much. [13:23:38.0394] Kris Kowal: 20 minutes for the Module Harmony discussion seems wildly optimistic [13:23:47.0824] looking forward to it though [13:40:39.0258] Hey all, we have an unusually large number of personal scheduling constraints for this meeting. This is making it hard to create a viable schedule. The chairs will be reaching out to those affected. If you are able to relax your constraints, please do so. It also helps if you can clarify where something is a *weak*preference vs a hard constraint by PRing the agenda. [14:45:40.0543] Just to confirm since it's my first time attending: this channel will have a Jitsi URL posted somewhere once the meeting starts tomorrow? [14:46:00.0042] Andrew Brown: there's a sign-up form for the call link [14:46:17.0351] Can you point me to it? [14:50:01.0273] you're right, there is indeed no link yet in the reflector [14:50:15.0144] we will send a link to the sign-up form before the meeting [15:03:38.0111] draft schedule is up! link in the reflector. 2022-11-29 [00:26:47.0914] Are we getting the video conference signup link? [00:27:49.0101] I take it the meeting starts in half an hour. I don’t see it on the Reflector [00:40:38.0502] It was posted on the Reflector 30mins ago. Please refresh. [00:44:11.0167] We are now admitting people to the Google Meet. Please use Incognito/Private browsing mode - because that will allow you to specify your name. So you can call yourself "Name (Affiliation)". [00:59:00.0140] We will begin the meeting in one minute! [00:59:21.0130] If anyone has any troubles dialling in, please say here. [01:00:10.0351] Is tcq not working for anyone else? [01:00:14.0457] I get meeting not available [01:00:33.0478] try reloading the page - that worked for me yulia [01:01:34.0132] I still get meeting not found. should it be TKTK or is the meeting id different? [01:01:45.0693] The signin form is available on the Reflector - please can all in-person attendees fill it out too: https://github.com/tc39/Reflector/issues/446 [01:02:41.0463] I just fixed the TCQ link on the draft schedule [01:02:50.0593] hm, maybe I'd opened it as a fresh link from the Reflector issue [01:03:00.0879] is it an obligation for the host to have a second CoC in play? obviously not a problem if they do have one, but somehow I don't have a recollection of that before [01:03:04.0405] Found it -- the link is different than the text [01:03:18.0271] Is it viable to turn off the lights above the projection screen? [01:03:28.0929] Are those lights independent? [01:03:40.0980] * is it an obligation for the host to have a second CoC in play? obviously not a problem if they do have one, but somehow I don't have a recollection of that before [01:05:38.0645] Romulo Cintra: Can you help with lights? [01:07:01.0825] Thank you Jason and Ashley and Kevin for volunteering for notes! [01:07:20.0538] btw the bot is using a new language model, which hopefully is better [01:07:22.0404] we'll see though [01:07:28.0603] > <@robpalme:matrix.org> Romulo Cintra: Can you help with lights? ? [01:08:06.0157] Romulo Cintra: Thx! [01:08:06.0852] Thank you [01:08:07.0861] That's much better [01:08:22.0269] The room is now is dark mode, which helps us to see the projector screen. [01:09:03.0415] 🌓 [01:09:11.0385] Why is the "attendees" section in the notes only for remote people? Having the abbreviations there helps also for in-person attendees [01:09:23.0534] it should be for both [01:10:16.0214] Please add your name, abbreviation and organization to the attendees section of the notes (each day). [01:12:06.0207] Rob Palmer: I can't find the sign in form for remote attendees [01:12:17.0823] Only the in person form [01:12:20.0710] Justin Ridgewell: it's on the reflector [01:12:24.0321] please refresh [01:12:37.0698] I did... [01:12:41.0617] it was added a short while ago, you might have the old page loaded [01:13:02.0412] It's in the video link section [01:13:22.0226] Ahh, got it [01:13:30.0800] It's tooooooo early. [01:14:04.0252] BTW TCQ seems to still be on the Welcome/Intro [01:14:52.0711] ryzokuken [01:15:06.0217] Thanks [01:16:32.0661] > <@jridgewell:matrix.org> It's tooooooo early. don't you mean late [01:16:57.0796] 4am in NYC. I just woke up [01:17:04.0724] ohhh right whoops [01:17:22.0228] true dedication [01:17:40.0359] > <@jridgewell:matrix.org> 4am in NYC. I just woke up 1am in California! [01:17:55.0750] i went to sleep at 8:30 [01:17:56.0213] Thank you for your dedication! [01:18:21.0329] > <@shuyuguo:matrix.org> i went to sleep at 8:30 That is about when I went to sleep. [01:18:25.0986] i don't want to be dedicated i want to be asleep [01:20:22.0963] > <@shuyuguo:matrix.org> i don't want to be dedicated i want to be asleep eternal mood [01:20:40.0763] except that it's 6pm in Tokyo for this meeting [01:20:50.0766] I'm now doomed for US meetings instead lol [01:28:34.0991] Slides: https://docs.google.com/presentation/d/1sWm_ynPyH_4XHUVwRuGIB6McALQHZXdENncu7BD1q9Y/ [01:42:46.0509] If anyone struggles with screen-sharing, please just post a link to your slides here and someone helpful can screenshare on your behalf. [01:47:10.0309] > <@rkirsling:matrix.org> is it an obligation for the host to have a second CoC in play? obviously not a problem if they do have one, but somehow I don't have a recollection of that before this was a non-rhetorical question re: hosting, btw [01:48:19.0575] Is there a list of in-person attendees that the remote folks can see? [01:48:31.0819] I can't tell who is in the room at the moment [01:49:12.0029] Waldermar, I'll update the notes doc to ensure in-person attendees are listed [01:50:14.0648] I think we actually skipped the "traditional" in-person introductions at the start of the meeting, though that's understandable given how packed the agenda is. [01:51:15.0044] we did a quick names-only runaround in the room before the meeting started - I will ensure this goes in the notes [01:53:42.0548] waldemar: we have added "in-person" and "remote" labels to the attendee list in the notes [01:55:38.0344] > <@rkirsling:matrix.org> is it an obligation for the host to have a second CoC in play? obviously not a problem if they do have one, but somehow I don't have a recollection of that before * this was a non-rhetorical question re: hosting, btw (am I just forgetful/unobservant?) [01:56:19.0970] > <@rkirsling:matrix.org> this was a non-rhetorical question re: hosting, btw (am I just forgetful/unobservant?) apologies if you were taken by surprise, I tried to give a heads up in the reflector post [01:56:51.0052] no need for apology, just was curious about it [01:56:54.0673] this was basically a request by the Igalia CoC committee and the chair group decided to respect it, I'm not sure if a second CoC has been applied in the past. [02:13:23.0195] What does STR mean? [02:13:35.0545] it was answered in #tc39-beginners:matrix.org [02:13:40.0676] Steps to reproduce [02:13:50.0239] Oh thanks [02:14:05.0000] someone want to PR the glossary with that? seems useful [02:18:26.0580] > <@bakkot:matrix.org> someone want to PR the glossary with that? seems useful ill do it [02:20:43.0998] is it used enough to justify? seems like we all shared the same confusion, so it might not be encourageable? [02:20:52.0544] * is it used enough to justify? seems like we all shared the same confusion, so it might not be encourageable? [02:21:31.0256] We often say “test case” for this sort of thing [02:27:48.0410] https://github.com/tc39/how-we-work/pull/124 <- PR, if we want to merge it. [02:30:20.0947] “groupedBy” [02:34:57.0568] Three options or two? [02:35:08.0670] A static method results in a pretty bad developer experience though. I've had some of the same issues with some of the Temporal APIs [02:35:44.0532] which do you mean? [02:35:49.0591] https://github.com/tc39/how-we-work/blob/main/presenting.md#temperature-checks [02:37:37.0472] I discussed this awhile back about the poor DX around things like `Temporal.Duration.compare` being static only, though I haven't followed up. Until something like pipeline advances, it breaks left-to-right reading order. [02:37:44.0490] tbh I feel like this is about more than just the method at hand is the thing [02:38:19.0875] given the difficulty of adding methods to Array in general [02:38:54.0529] I agree there, its disheartening that `Array` is constantly so difficult to extend. [02:39:05.0424] should we consider just pursuing this as an iterator helper method? [02:40:54.0963] Its unfortunate because I believe this is something that belongs on Array.prototype. And this won't be the last time we run into this. [02:43:51.0620] What is it that the static methods is consistent with? Object groupBy? [02:43:57.0409] Another way to approach this might have been a `.collect(collector)` a. la. Java collectors (https://docs.oracle.com/javase/8/docs/api/java/util/stream/Collectors.html), but that's a completely different proposal [02:44:38.0590] (i.e., `array.collect(Collector.groupBy(keySelector))`) [02:44:47.0120] > <@rpamely:matrix.org> What is it that the static methods is consistent with? Object groupBy? Object.fromEntries, which produces an object given an array [02:45:01.0169] Oh I see ok [02:45:49.0851] I see that as quite a different helper from groupBy [02:46:12.0038] did i miss the bit where we had bathroom explainations? cc ryzokuken [02:46:16.0019] I like the suggestion for `Object.groupBy` and `Map.groupBy`, where you select the return type with the constructor. [02:46:24.0655] > <@yulia:mozilla.org> did i miss the bit where we had bathroom explainations? cc ryzokuken oops, sorry [02:46:25.0127] "Entries" are an array representation of an object. groupBy is an operation on an array. [02:46:42.0290] I could do that right after this item [02:46:50.0022] > <@rpamely:matrix.org> "Entries" are an array representation of an object. groupBy is an operation on an array. couldn't the same argument be made about entries? [02:47:09.0528] Entries have to have a specific shape though [02:47:43.0387] groupBy is more like reduce than fromEntries [02:47:48.0643] > <@kriskowal:matrix.org> I like the suggestion for `Object.groupBy` and `Map.groupBy`, where you select the return type with the constructor. I'm not entirely against it, but I'd again rather see something like a formal collector API that those statics could be built off of. [02:47:51.0246] > <@rpamely:matrix.org> "Entries" are an array representation of an object. groupBy is an operation on an array. Of a map (and by implication objects-as-dictionaries, which are map-alike) [02:48:32.0148] > <@rbuckton:matrix.org> I'm not entirely against it, but I'd again rather see something like a formal collector API that those statics could be built off of. Generators-as-collectors come to mind. [02:48:32.0251] > <@kriskowal:matrix.org> Of a map (and by implication objects-as-dictionaries, which are map-alike) Yes I guess specifically Entries are a representation of key-value pairs, which objects and Maps can represent [02:48:46.0781] > <@kriskowal:matrix.org> I like the suggestion for `Object.groupBy` and `Map.groupBy`, where you select the return type with the constructor. My understanding is this proposal only adds a static to Array right? Is there an addition to Object and Map also and i missed that? [02:48:58.0933] > <@kriskowal:matrix.org> Of a map (and by implication objects-as-dictionaries, which are map-alike) what is this "of" attached to? [02:49:18.0927] > <@jasew:matrix.org> My understanding is this proposal only adds a static to Array right? Is there an addition to Object and Map also and i missed that? I think it would only add statics to Object and Map, not to Array at all [02:49:46.0110] I see [02:50:03.0246] Given users are used to finding helpers like `reduce` on the array prototype, I don't think it will be obvious to look for these on Object and Map [02:50:03.0759] Yes, I believe yulia suggested using Object and Map constructors as receivers _instead of_ Array. [02:50:16.0462] yeah i think that was just a mistake in the slides [02:51:38.0919] Yes, it was a typo to have `Array.groupBy` and `Tuple.groupBy` [02:51:43.0461] In any case, `X.of` is the good precedent for `X.groupBy`. Both return instances of `X` and have an open range for values of `X`. [02:51:59.0132] It should just be `Object.groupBy`, `Map.groupBy` (and `WeakMap`), and possibly `Record.groupBy` [02:52:17.0762] That is, provides clear guidance for a library `SortedMap.groupBy` [02:52:18.0285] This makes more sense to me now, thanks for clarifying [02:52:24.0871] * This makes more sense to me now, thanks for clarifying [02:52:55.0334] > <@rpamely:matrix.org> Given users are used to finding helpers like `reduce` on the array prototype, I don't think it will be obvious to look for these on Object and Map Object.entries can be created from a map. There we don't have entriesToObject on the map prototype or array prototype, so I think we have precedent for this. That said, i do understand where you are coming from, I am just agreeing with eemeli that we have similar patterns already [02:53:24.0027] `groupBy` is kind of a weird name if this is a static [02:53:30.0489] though I don't have a better one [02:54:35.0546] fromGrouped? like fromEntries? I don't have a better name either [02:54:53.0506] > <@jasew:matrix.org> fromGrouped? like fromEntries? I don't have a better name either The input is not grouped [02:55:01.0014] The Java grouping collector is `groupingBy`, which was one of the options listed as an alternative for the .prototype version, if we're going to change the name anyways [02:55:30.0163] if we're ok with `groupingBy` it can probably still be on Array.prototype [02:55:41.0776] I doubt that will have conflicts like `group` and `groupBy` did [02:55:55.0449] > <@bakkot:matrix.org> if we're ok with `groupingBy` it can probably still be on Array.prototype That would be my preference, though I'd lean more towards `.groupedBy` in that case. [02:56:03.0708] though `groupingByToMap` is... not... a great name [02:56:21.0741] yeah I think I like `groupedBy` personally [02:56:23.0978] if we continue to go with the prototype, then I don't think we can again just implement it and hope for bugs if there are problems [02:56:33.0365] is there a more robust way we can test this name? [02:56:36.0629] yulia: even with a really weird name like `groupingBy`? [02:57:01.0758] `groupingBy` indicates something that produces a thing will do the grouping (i.e., Java's `Collectors.groupingBy`), where as `groupedBy` indicates something that produces the grouped result. [02:57:07.0829] conflicts with `group` and `groupBy` are not too surprising because those are natural names that people might use [02:57:10.0092] groupingBy and groupToObject is probably safe... but is this really better than it being a static if we are looking for really weird anmes? [02:57:11.0166] Object.groupsFromArray? [02:57:27.0716] > but is this really better than it being a static eh, personally I think so [02:57:36.0328] Object.toCategories? [02:57:49.0383] idk [02:57:51.0684] statics are quite strange for this [02:57:51.0909] Something with `from` in the name often implies one type (input) to another (usually the same as the reciever) [02:57:56.0361] > <@yulia:mozilla.org> groupingBy and groupToObject is probably safe... but is this really better than it being a static if we are looking for really weird anmes? One concern with static is chaining too. Without pipeline you lose the ability to easily chain at the end of array operations. [02:58:20.0847] > <@rpamely:matrix.org> One concern with static is chaining too. Without pipeline you lose the ability to easily chain at the end of array operations. you would not be chaining array methods anyway [02:58:31.0337] sure you would [02:58:32.0534] > <@yulia:mozilla.org> Object.toCategories? the FP community will surely have our heads 😂 [02:58:34.0012] As I mentioned earlier, I wonder if a comprehensive check for a more general-purpose `Array.prototype.collect` method would make sense, plus a formalized Collector API. [02:58:45.0742] `x.filter().groupBy()` is a pretty common thing to want [02:59:00.0457] but you would have an object after groupBy [02:59:00.0795] `Object.groupBy(arr.filter().map())` vs `arr.filter().map().groupedBy()` [02:59:06.0116] > <@bakkot:matrix.org> `x.filter().groupBy()` is a pretty common thing to want Yes, I use this pattern all the time with my own libraries. [02:59:07.0535] ah i see what you folks mean [02:59:29.0165] thats fair [02:59:38.0299] `groop` [02:59:45.0801] > <@rbuckton:matrix.org> As I mentioned earlier, I wonder if a comprehensive check for a more general-purpose `Array.prototype.collect` method would make sense, plus a formalized Collector API. I am not necessarily opposed to such a thing, but I have a hard time imagining it being as convenient as `groupBy`. `.collect(Collectors.groupBy)` is much worse than `groupedBy`. [03:00:09.0358] > <@rkirsling:matrix.org> `groop` it's called the JS language for a reason [03:00:36.0122] i mean, we can't even decide on a name for the language so ;) [03:00:51.0403] officially its ECMAScript, which is the only still-capitalized thing for Ecma international [03:01:13.0274] > <@yulia:mozilla.org> but you would have an object after groupBy That depends on the implementation. A query/sequence-like `groupBy` would produce an array of groupings (i.e., key/values pairs) [03:01:13.0628] Using `groupBy` will be good for discovery. [03:01:18.0447] so i prefer to think that its still European Computer Manufacturer's Association Script, or, per the sporting world, soccer script [03:01:25.0508] ill show myself out [03:01:44.0575] has anyone ever tried to align the capitalization of ECMAScript? haha [03:02:10.0138] clearly a task worthy of time and effort 🙄 [03:02:11.0508] you know that's not the OG ecma [03:02:13.0481] > <@kriskowal:matrix.org> Using `groupBy` will be good for discovery. In what sense? Searching docs? [03:02:16.0966] the OG ecma is carton manufacturers [03:02:21.0993] https://www.ecma.org/ [03:02:42.0960] > <@rpamely:matrix.org> In what sense? Searching docs? Googling “mdn groupBy” on a lark and finding useful results. [03:02:43.0399] > <@bakkot:matrix.org> I am not necessarily opposed to such a thing, but I have a hard time imagining it being as convenient as `groupBy`. `.collect(Collectors.groupBy)` is much worse than `groupedBy`. Yes, but if we're considering statics anyways, a `.collect` would preserve LTR reading/method chaining [03:03:10.0380] i.e., `.collect(Map.groupingBy)` for example [03:03:31.0315] > <@kriskowal:matrix.org> Googling “mdn groupBy” on a lark and finding useful results. True, but "mdn group array" is a likely search too [03:03:37.0343] well, `.collect(Map,groupingBy, fn)`, really [03:03:51.0174] where any solution would be found hopefully.. [03:03:56.0199] > <@shuyuguo:matrix.org> the OG ecma is carton manufacturers Carton: the language formerly known as variously ECMAScript and JavaScript. [03:04:10.0907] > <@bakkot:matrix.org> well, `.collect(Map,groupingBy, fn)`, really Well, if you follow the Java design, `.collect(Map.groupingBy(fn))` [03:04:25.0096] > <@kriskowal:matrix.org> Googling “mdn groupBy” on a lark and finding useful results. no matter what option we do "groupBy" should be mentioned in the docs [03:04:29.0680] > <@kriskowal:matrix.org> Carton: the language formerly known as variously ECMAScript and JavaScript. That's actually kinda catchy [03:04:32.0473] These full on Java suggestions are hurting my eyes [03:04:44.0618] re: discoverability, I'm more worried about people being able to do `x.gr...` in the console and get autocomplete which suggests the thing they wanted [03:04:47.0240] We will resume the meeting in 56 mins at 13:00 local time. [03:04:50.0399] * re: discoverability, I'm more worried about people being able to do `x.gr...` in the console and get autocomplete which suggests the thing they wanted [03:04:55.0159] which only works for proto methods, not static [03:04:59.0531] > <@bakkot:matrix.org> well, `.collect(Map,groupingBy, fn)`, really `.collect(Map, fn)` would also work, with a collectors protocol [03:08:42.0903] > <@kriskowal:matrix.org> Carton: the language formerly known as variously ECMAScript and JavaScript. cool devs in 2030 only use Carbon and Carton [03:09:15.0805] Are we changing our mind on static methods? [03:09:38.0399] different people always had different opinions I think [03:09:42.0240] I'm just kinda exhausted with seeing naming conflicts with the ecosystem. [03:09:49.0515] I don't have that firm of an opinion at the moment [03:09:50.0826] `group` was impossible to query for [03:09:51.0998] need to think about it more [03:10:20.0593] We could seriously consider something like F#'s computation expressions (something jschoi was looking into) to create query DSLs: ```js // computation expression example using decorators and do-expression syntax: const array = ...; const result = @query do { for x in array groupBy x.name into y select y }; ``` [03:10:33.0137] * We could seriously consider something like F#'s computation expressions (something jschoi was looking into) to create query DSLs: ```js // computation expression example using decorators and do-expression syntax: const array = ...; const result = @query do { for x in array groupBy x.name into y select y }; ``` [03:10:48.0734] > <@jridgewell:matrix.org> I'm just kinda exhausted with seeing naming conflicts with the ecosystem. this is what I'm saying; what I want is a solution that we can also apply to the next time this happens [03:11:19.0213] > <@rbuckton:matrix.org> We could seriously consider something like F#'s computation expressions (something jschoi was looking into) to create query DSLs: > > ```js > // computation expression example using decorators and do-expression syntax: > const array = ...; > const result = @query do { > for x in array > groupBy x.name into y > select y > }; > ``` that is too many things [03:11:23.0359] JS does not have room for that [03:11:26.0114] * this language does not have room for that [03:11:29.0021] * JS does not have room for that [03:12:55.0466] if any of us had large twitter followings we could poll the community on `Object.groupBy` vs `[].groupedBy` [03:13:13.0152] just don't put a joking "break the web" option in or people will take it seriously and get mad that we don't [03:13:42.0138] > <@bakkot:matrix.org> if any of us had large twitter followings we could poll the community on `Object.groupBy` vs `[].groupedBy` I'm less convinced of the efficacy of twitter polls given the mass exodus of many in the tech community off of that platform and onto platforms like Mastodon. [03:14:40.0015] > <@rbuckton:matrix.org> I'm less convinced of the efficacy of twitter polls given the mass exodus of many in the tech community off of that platform and onto platforms like Mastodon. Or, as we like to call it, the Lemmer Web. [03:15:02.0367] (for Christine Lemmer-Webber, one of the authors of ActivityPub) [03:15:11.0955] > <@bakkot:matrix.org> JS does not have room for that I'm not so sure. `async do {}` is not so far off from what `@async do {}` might be in terms of how `async {}` works in F#. [03:15:37.0749] But even if it is viable, its a long way off. [03:17:28.0932] > <@rbuckton:matrix.org> I'm less convinced of the efficacy of twitter polls given the mass exodus of many in the tech community off of that platform and onto platforms like Mastodon. eh, well, you can run it and see what results you get [03:17:43.0911] seems like there are still enough people around to get a decent sample size [03:18:06.0817] We need scoped prototype methods [03:18:17.0012] if we could get decent sample size in the first place, anyway [03:18:42.0546] ```js { "use groupBy"; [].groupBy; // => function } [].groupBy; // => undefined ``` [03:19:00.0397] > <@jridgewell:matrix.org> We need scoped prototype methods Jake was talking about this too, but how would it work? [03:19:02.0411] > <@jridgewell:matrix.org> We need scoped prototype methods You mean like https://github.com/tc39/proposal-extensions? [03:19:31.0508] Similar, but access would still use `.` instead of `::` [03:19:34.0201] > <@bakkot:matrix.org> if we could get decent sample size in the first place, anyway Anyway we can do a parallel Mastodon poll! [03:19:37.0545] Else it's just a pipe. [03:19:56.0391] > <@jridgewell:matrix.org> Else it's just a pipe. That is the way [03:20:55.0095] If only we could do something like C#'s extension methods, but that requires runtime typing [03:21:51.0608] > <@littledan:matrix.org> Jake was talking about this too, but how would it work? Roughly, don't we already have similar tracking for strict property access? [03:22:04.0879] Just need to track prototype scopes now, for every property access... [03:22:20.0538] /me runs away [03:26:40.0726]