TC39 Delegates
2025-07-28
plaintext logs
prev next
source on github
16:44
<Rob Palmer>
Hello all
16:45
<Rob Palmer>
We've posted a new TCQ Reloaded link on the Reflector. Please can folk verify they have access. https://github.com/tc39/Reflector/issues/565 (and do NOT post the TCQ link here)
16:48
<Jesse>
works for me!
16:50
<Rob Palmer>
Plenary begins in 9 minutes!
16:54
<Jesse>
looks like the privs for the notes google doc may need to be changed (normally it's available for edits right away, but I just found it to be read-only)
16:55
<Rob Palmer>
permissions on the docs are fixed
16:56
<Jesse>
can confirm
17:06
<Michael Ficarra>
FYI the note taker has newlines
17:06
<Michael Ficarra>
the transcriptionist I mean
17:06
<Chris de Almeida>
new TCQ is not a complete rewrite. it is a fork, main changes were to add abstraction to use with different infra, I think it's made containerized now. shouldn't be much, if anything, in the way of feature diffs
17:06
<Chris de Almeida>
https://github.com/zalari/tcq/compare/bterlson%3Atcq%3Amaster...reloaded
17:07
<Chris de Almeida>
this should save us from some pain though: https://github.com/bterlson/tcq/commit/6039a03d7c602f82af2203acc0e938cb59dd3a84
17:08
<ljharb>
the notetaker also double spaces after periods
17:09
<Aki>
regular expressions hide all sins
17:10
<ljharb>
not the irregular ones
17:13
<snek>
only commented out in the html... 😈
17:13
<nicolo-ribaudo>
Writing a different number of spaces each time so that a regexp cannot catch me
17:13
<Chris de Almeida>
get's the job done 😄
17:13
<Chris de Almeida>
the bad-linebreaks script will catch you
17:14
<Aki>
s/ / /g
17:14
<ljharb>
repeated replace-alls of . to . will get you eventually
17:15
<Justin Ridgewell>
Regex has the power
17:15
<Chris de Almeida>
const reBadLinebreaks = /(?<=[\w\d \p{P}])\n(?=[\w\d ])/gu;
const reExtraWhitespaceParagraph = /^ +| (?= )| +$/gm;
const reExtraWhitespaceList = /(?<=^ {0,}[-*+] |\d+\. ) +|(?<=\w+ ) +| +$/gm;
17:17
<ljharb>
(consensys is metamask, in case you're like me and have never heard of the former)
17:17
<naugtur>
I'm always hesitating which one to put as company
17:27
<waldemar>
Curious to hear about the exciting new features in ECMA 404…
17:28
<ryzokuken>
"...features not found"
17:28
ryzokuken
🏃
17:41
<Chris de Almeida>
https://github.com/tc39/code-of-conduct/issues/62
17:47
<Michael Ficarra>
I will start including a summary as my final slide
17:57
<naugtur>
And we could edit it on the spot as conclusions are reached.
17:57
<Michael Ficarra>
thank you very much for the presentation, this helped clarify what was expected from summary and conclusions
17:58
<Aki>
oh yay, i'm so happy to read that Michael Ficarra
17:59
<saminahusain>
Thank you for the feedback.
17:59
<keith_miller>
https://github.com/tc39/ecma262/issues/3652
18:02
<nicolo-ribaudo>
naugtur It might be useful to screen-share https://tc39.es/ecma262/#sec-setfunctionname for Mark
18:11
<nicolo-ribaudo>
And highlight step 5.b.i :)
18:12
<naugtur>
Sorry, struggled to find where this notification came from
18:16
<eemeli>
An explicit Annex B list seems like the best solution here.
18:19
<nicolo-ribaudo>
Chairs, if we end up having 20 minutes free before lunch I'm happy to do my 20m topic
18:20
<nicolo-ribaudo>
Oh well they are both 20m, I guess any of them
18:38
<ljharb>
um, i had the reloaded queue open and never saw any emoji
18:40
<naugtur>
nicolo-ribaudo: I'm wondering if this change could have impact on require(esm) in Node. Probably not as it's not supporting top level await
18:40
<eemeli>
Heh, I saw the emoji but they went away when I switched out of my browser on Android in order to figure out what they meant.
18:44
<nicolo-ribaudo>
Yeah this is only about dynamic import of TLA module graphs
18:48
<naugtur>
I feel like we should be seeing something else in screen share
20:07
<Michael Ficarra>
@Chris de Almeida DJM is @Dmitry Makhnev
20:07
<Michael Ficarra>
DJM and DLM both gave +1
20:07
<Chris de Almeida>
yep
20:07
<Chris de Almeida>
but DLM was incorrect in there as DM until I fixed it 🙂
20:12
<TabAtkins>
So in favor of this, it's great
20:31
<Andreu Botella>
I thought this would be a first step, to then consider doing the same for set iterators and so on
20:32
<Michael Ficarra>
@Andreu Botella true, giving up on freezing and instead exploring the Array special-casing is ignoring how we could apply the freezing to other built-in iterators
20:33
<nicolo-ribaudo>
Chris de Almeida: I need to stop taking notes at the top of the hour, rather than when my topic starts as I said before
20:35
<Michael Ficarra>
spreading array literals happens more than you might think
20:35
<ljharb>
i would be surprised if arrays weren't many 9's of the things being spreaded
20:36
<Michael Ficarra>
we don't have an "optionally include this element" so it's common to spread ...(a ? [b] : []) within an array initialiser
20:36
<kriskowal>
Our position at Agoric regarding initially immutable intrinsics is really subtle. When we create the HardenedJS environment, we don’t just freeze all of (what we call) Shared Intrinsics, but some of those intrinsics have to be changed, like Date.now, Math.random, so in general we rely on the pervasive mutability of the language to enable language evolution. We do not believe the specific proposed change hinders us in building HardenedJS environments. But, for the evolution of the web, “Even the very wise do not know all ends” and pervasive mutability has been good actually, allowing us to plug all manner of holes.
20:46
<Michael Ficarra>
@keith_miller translate it back?
20:47
<ljharb>
see translationparty.com
20:48
<Aki>
ISO policy would theoretically cover Discourse. ISO's whole point (imo) boils down to "use LLMs all you want, just don't show your output to anyone"
20:50
<nicolo-ribaudo>
We can redact the notes before publishing them
20:50
<Michael Ficarra>
we saw that other venues are covering this behaviour in their CoC-equivalent policies, why wouldn't we?
20:52
<ljharb>
is the ISO's in a code of conduct? that's a very specific kind of document.
20:54
<Chris de Almeida>
I don't think so
20:54
<Michael Ficarra>
I think the ISO and ACM policies predate people calling these policies out as "CoC"
20:54
<Michael Ficarra>
but that's effectively what they are
20:54
<Michael Ficarra>
they govern your behaviour in your various interactions with those groups
20:54
<Chris de Almeida>
https://www.iso.org/publication/PUB100011.html
20:56
<ljharb>
so they have a coc, and the ai policy is in a different document
20:58
<kriskowal>
I have learned that people are avoiding em-dashes and Oxford commas in order for their manually-intelligent writing less resemble AI-generated content. Is their a sufficiently reliable criterion that distinguishes rude vs non-rude ~generative~ content, regardless of whether it’s generated?
20:58
<ljharb>
psh, you can pry my oxford comma from my cold dead hands
20:58
<Michael Ficarra>
you prefer having more than one document that governs committee interactions?
20:58
<kriskowal>
You have failed CAPTCHA.
20:58
<ljharb>
we already do. that's what "how we work" is, and the contribution guide, and a number of other things
20:59
<Michael Ficarra>
none of those warn of formal consequences if not followed
20:59
<ljharb>
nor need they, we can impose consequences any time we want, arbitrarily
20:59
<ljharb>
i don't need permission to hide someone's comments on my proposal repo, for example
21:00
<Michael Ficarra>
we should not do that, we should have a written policy of the things that will have formal consequences associated with them
21:00
<Michael Ficarra>
like a CoC
21:00
<ljharb>
the purpose of this addition, wherever it lives, is to establish a norm that we can cite; it doesn't imbue us with any new capabilities
21:01
<Michael Ficarra>
somebody interacting with committee should have all the tools/info they need to make a contribution without worry that it will come with consequences
21:01
<Michael Ficarra>
such as by reading the CoC
21:01
<ljharb>
ok but that's a "should" you believe, it's not the current rule/policy, nor is it the case in basically any venue
21:01
<ljharb>
and the CoC is already not the sole place one must read to gather that info
21:02
<ryzokuken>
if the idea is to moderate contributions from external folks on github and discourse i don't see why formally making it a part of our coc is necessary
21:02
<Michael Ficarra>
it's not?
21:02
<ljharb>
no. the contributor guide is the primary place, and that links to the CoC
21:02
<Andreu Botella>
I can take over
21:02
<ljharb>
the CoC talks about comportment and demeanor. the contributor guide talks about how you contribute.
21:02
<Michael Ficarra>
violations of the contributor guide don't get you banned
21:02
<ljharb>
they absolutely could if they're repeated
21:03
<ljharb>
and we wouldn't ban someone for a single AI-generated contribution anyways
21:03
<Michael Ficarra>
no matter how much I'd like to ban someone who omits the Oxford comma
21:03
<ljharb>
the goal isn't to ban people, it's to establish norms so enforcement isn't practically needed in the first place
21:04
<ljharb>
the consequences aren't the point - the explicit expectation is the point
21:04
<ljharb>
for example, if my kids only behave because of consequences then i've not done a good job :-) (which, to be clear, is empirically the case at times)
21:05
<Chris de Almeida>
to be clear, use of generative AI in a way that we have prohibited would 100% be subject to enforcement action via the CoC regardless of where the guidance/rules live
21:14
<Michael Ficarra>
where will this bikeshedding be happening? the same coc repo thread?
21:15
<Chris de Almeida>
that issue and the forthcoming PR with the proposed text
21:28
<naugtur>
nicolo-ribaudo: part of your long paragraph is really mixed up. couldn't save it
21:28
<nicolo-ribaudo>
I'll go through it, don't worry
21:28
<nicolo-ribaudo>
Thanks for the ping
21:46
<Andreu Botella>
let me fix my mic
21:52
<nicolo-ribaudo>
The whole presentation is already a summary :P
21:55
<nicolo-ribaudo>
Who volunteers to squeeze their 30min topic in the next 4 minutes?
22:00
<naugtur>
There were a couple moments where the transcriptionist would struggle to catch up and drop whole sentences. I think we could make a conscious effort to make a ~1 second pause after each outburst of content when one person is talking, especially with longer sentences where logic of the sentence is non-trivial.