Listen Notes: il miglior motore di ricerca per podcast
Episode 312: The C++26 Contracts Fiasco

Trascrizione completa di Episode 312: The C++26 Contracts Fiasco

Download transcripts: JSON SRT VTT HTML

TRASCRIZIONE 🔗

WEBVTT 00:00:00.160 --> 00:00:02.000 And so it had very high consensus. 00:00:02.000 --> 00:00:17.280 So, I mean, this still leaves 25% of people that are like upset about it, but we have this process where everyone comes to the meetings, we have all the points of view in the room at the same time, everyone gets to have their say, sometimes for way too long, and then finally we come to a vote, and people are persuaded or they're not persuaded. 00:00:17.280 --> 00:00:24.480 The idea that we missed something, there was an oopsie, and the design has defects in it that no one realized before is just incorrect. 00:00:24.480 --> 00:00:29.920 All of the things that have been said since the CD and since the disc went out were said before either one of them went out. 00:00:30.239 --> 00:00:30.320 Right. 00:00:30.320 --> 00:00:32.479 There has not been substantial new information. 00:00:32.479 --> 00:00:36.159 That's one of the reasons why I point to this as a fundamental breakdown in the process. 00:00:36.399 --> 00:00:48.560 What it feels like to me is happening is people that were salty about it, who have a lot of pull in the community, went and lobbied with people that they thought they would get a friendly year with on those national bodies, and they just changed people's minds. 00:00:48.560 --> 00:00:50.000 That's what it feels like happened to me. 00:00:50.000 --> 00:01:00.560 I don't have any proof of that, but that that's I mean, when you see no technical difference and yet you see the votes change radically, like, well, people's opinions changed, but the technical details didn't change, so what's going on there? 00:01:10.079 --> 00:01:16.159 Welcome to ADSB the podcast, episode 312, recorded on October 28th, 2026. 00:01:16.159 --> 00:01:22.480 My name is Connor, and today with my co-host Bryce, we chat with Zach Lane for part two of this three-part conversation. 00:01:22.480 --> 00:01:37.760 In this episode, we chat about the drama in the C ISO committee relating to C26 contracts and how the DIS draft international standard for C26 failed when it went to a vote. 00:01:53.373 --> 00:01:55.694 So uh is this it? 00:01:55.694 --> 00:01:57.134 Is this the end? 00:01:57.453 --> 00:01:58.093 The drama. 00:01:58.093 --> 00:01:58.974 The drama. 00:01:58.974 --> 00:02:09.280 I had to mess up the rice because like I I guess full uh full disclosure, not full disclosure, for potentially you know, I mean I haven't been the last one. 00:02:12.080 --> 00:02:22.960 Right, I mean I have the I love I love the time at the Fourth meeting, but then like I skipped a meeting last year. 00:02:22.960 --> 00:02:27.680 I was actually sadly at the to find the ones. 00:02:27.680 --> 00:02:32.080 was like the 2025 Kona meeting. 00:02:32.080 --> 00:02:41.199 And I was not able to play that movie because I may have had a is your wedding! It was your wedding! It wasn't twenty twenty five, it must have been twenty twenty six. 00:02:41.520 --> 00:02:44.719 Wait, uh the Kona Kona meetings are usually beginning of the year. 00:02:44.719 --> 00:02:45.599 My wedding was in November. 00:02:45.759 --> 00:02:48.080 It was in it was No, this was this was one in November. 00:02:48.080 --> 00:02:48.719 Yeah, yeah. 00:02:48.719 --> 00:02:50.080 Yeah. 00:02:50.319 --> 00:02:50.960 Okay, you were in the middle of the year. 00:02:51.120 --> 00:02:55.199 Yeah, it was that was my ten my ten year my ten year anniversary on the committee. 00:02:55.199 --> 00:03:01.919 It would have been the perfect meeting for me to like end close that that that that chapter, but no, I I had to go anyways. 00:03:03.280 --> 00:03:05.759 This meeting coming up is I think my ten-year anniversary, now that I think about it. 00:03:05.759 --> 00:03:06.960 I didn't even think about that to right now. 00:03:07.199 --> 00:03:08.000 Did you join after this one? 00:03:08.000 --> 00:03:09.439 Did you join after me then? 00:03:09.680 --> 00:03:10.240 Yeah, yeah. 00:03:10.240 --> 00:03:12.800 I think uh a year or two after you, maybe longer. 00:03:12.800 --> 00:03:13.280 Yeah. 00:03:13.280 --> 00:03:24.560 You were you were the one that gave me my orientation to the first meeting I went to, and that you gave me advice like I continue to give to brand new attendees, which is go to every room and see what they're like because you're not gonna believe how different they are from each other. 00:03:24.560 --> 00:03:28.719 And sure enough, at the end of the week, sometimes at such bases, people were like, so were they different? 00:03:28.719 --> 00:03:30.479 And they were like, oh my god, they were so different. 00:03:30.479 --> 00:03:33.039 It's like not even like it's the same organization. 00:03:33.360 --> 00:03:36.879 I I don't I do not think I will be at another C committee meeting. 00:03:36.879 --> 00:03:37.439 Okay. 00:03:37.439 --> 00:03:38.800 Probably probably ever. 00:03:39.039 --> 00:03:48.960 Alright, well, so he so you you I was trying to give like a little rapid fire update on everyone's like committee involvement, and then Bryce been on a little tangent, which has never happened before, surprisingly. 00:03:48.960 --> 00:03:50.639 Uh that's a unique experience. 00:03:50.639 --> 00:03:53.520 My point is I haven't been involved in years. 00:03:53.520 --> 00:03:57.759 Bryce has been going, but stepped down as Luke chair a couple years ago. 00:03:58.080 --> 00:04:02.000 I'm I'm a backbencher now, which means I throw tomatoes at the uh at the leadership. 00:04:02.400 --> 00:04:06.159 And then I knew we were having this interview today, and I was like, is Zach still involved? 00:04:06.159 --> 00:04:07.199 Because I'm not involved, so I don't know. 00:04:07.199 --> 00:04:08.560 I don't I don't even read the papers. 00:04:08.560 --> 00:04:10.400 I don't see whose names are on the papers and whatnot. 00:04:10.400 --> 00:04:17.360 And then Bryce said, Yes, Zach is involved, and I was like, fantastic, because I do frequent the C subreddit from time to time. 00:04:17.360
Load full transcript
--> 00:04:17.759 Right. 00:04:17.759 --> 00:04:25.360 And uh I saw the dumpster fire that is the contracts mess slash C26 being. 00:04:25.360 --> 00:04:29.600 So I don't even set it up because I don't even know how to describe it. 00:04:29.600 --> 00:04:31.839 And then we'll get Zach to answer all our questions. 00:04:32.160 --> 00:04:47.759 So the the way that the standard actually gets made is the the committee produces a committee draft, and then we vote out that committee draft, and then that gets sent out to balloting for approval by the national bodies, which is all the various countries. 00:04:48.399 --> 00:04:51.120 So the CD goes out and they we get comments back, and we have to do that. 00:04:51.120 --> 00:04:52.240 Yeah, we get comments, that's right. 00:04:52.240 --> 00:04:56.240 And then there's another thing called a DIS that gets sent out, and that's the thing that gets voted on. 00:04:56.240 --> 00:04:56.399 Yeah. 00:04:56.800 --> 00:04:59.279 What is D D I S's draft international standard? 00:04:59.279 --> 00:05:00.959 The draft international standard, yes. 00:05:01.439 --> 00:05:06.160 And there's the the F DIS, which is the final draft international standard, which is like I think what they call it after the voting. 00:05:06.160 --> 00:05:09.519 I'm not sure about the F versus non-f DISC, but something like that. 00:05:09.759 --> 00:05:11.360 You gotta love processing committees. 00:05:11.519 --> 00:05:12.240 Uh yeah. 00:05:12.560 --> 00:05:19.920 So so first it goes out for, as Zach said, for the sort of like beta testing, like like you know, beta release where we get comments back on it. 00:05:19.920 --> 00:05:26.639 And then after that, we go and we send out, you know, we make some changes, and then we send out the final, the final draft version. 00:05:26.639 --> 00:05:32.639 We send out the committee's final draft version, hopefully, the disc ballot. 00:05:32.639 --> 00:05:42.399 And I don't think a disc ballot has failed before, although maybe No, the it has not ever failed before, and in fact, there's never been a no vote before. 00:05:42.399 --> 00:05:53.759 I thought that in the I think I thought that in eleven or oh three that the Irish NB like said that they were gonna no vote if they didn't it take out Autoputter, which was horribly broken. 00:05:54.000 --> 00:05:55.120 Yeah, but then it got taken out. 00:05:55.279 --> 00:05:55.439 Yeah. 00:05:55.439 --> 00:05:56.560 So they didn't no vote. 00:05:56.560 --> 00:06:04.000 So I don't think there's ever been a no vote vote before, and uh this time there were twelve, nine no votes. 00:06:04.000 --> 00:06:04.240 Yeah. 00:06:04.800 --> 00:06:06.000 There were twelve in favor, nine eight. 00:06:06.319 --> 00:06:14.319 One of the things that I did during my career on the committee was I worked very hard to recruit new national bodies. 00:06:14.319 --> 00:06:26.079 And the reason I wanted more national bodies on the committee was I did not like the idea of a single NB having of a single NB's no vote being enough to tank the standard. 00:06:26.079 --> 00:06:32.160 So if we had a lot more NB's, then we could have one or two no votes. 00:06:32.160 --> 00:06:40.079 So we wouldn't have this tyranny of like a single a single NB, which in some cases can be a single person being able to have a veto on the standard. 00:06:40.079 --> 00:06:46.959 And I'm very happy that we now have like enough NB's that that you know we we can sustain one or two or even three no votes. 00:06:46.959 --> 00:06:49.360 However, we had a we had a few more than that. 00:06:49.360 --> 00:07:16.720 And there's a whole separate drama about the the the US delegation abstained, but the US C plus plus committee, the US C committee wanted to vote to approve, but the US C committee has a parent organization that through a series of unfortunate events decided to overrule the uh US uh committee. 00:07:16.720 --> 00:07:18.959 So there's sort of two interesting things that happen here. 00:07:18.959 --> 00:07:22.800 One, there's like a very big procedural breakdown here. 00:07:22.800 --> 00:07:29.120 I think regardless of what you think about contracts, this never have sh should have gotten to this point. 00:07:29.120 --> 00:07:38.959 If there was this much sustained opposition, we, like the leaders of the committee, should have made sure that that it was addressed before it got to this stage. 00:07:38.959 --> 00:07:40.560 So clearly some rules were. 00:07:41.120 --> 00:07:41.920 there really wasn't, right? 00:07:41.920 --> 00:07:49.199 I mean we had we had we had lots of people that were very opposed to some of the aspects of contracts. 00:07:49.199 --> 00:07:54.079 Some for reasons that a lot of people I think identified with and thought were valid reasons. 00:07:54.079 --> 00:07:58.000 Some for reasons that seemed specious and were just like seemed rhetorical to me. 00:07:58.000 --> 00:08:01.839 But you know I don't really I don't really know a a lot about the contract stuff in detail. 00:08:01.839 --> 00:08:04.079 Like I I I'm I wasn't involved in those papers. 00:08:04.079 --> 00:08:12.480 I I will use contracts sometimes probably if it gets in you know if this if this gets resolved and we end up with contracts I'll probably use it a little bit but if we don't have it I'm not going to miss it. 00:08:12.480 --> 00:08:15.680 So don't don't think that this is some you know ideological thing for me. 00:08:15.680 --> 00:08:22.720 But um But ultimately, like there were people that were definitely opposed, and some of the reasons seem more reasonable than others. 00:08:22.720 --> 00:08:31.519 But what kept happening was every time we voted, the consensus grew until maybe at the end there was some noise in the consensus, maybe it went down a little bit, but it was always very, very high. 00:08:31.519 --> 00:08:38.399 We're talking about like 75%-ish of people in favor when usually consensus is considered two-thirds, right, not three-quarters. 00:08:38.399 -->
00:08:40.320 And so it had very high consensus. 00:08:40.320 --> 00:08:46.080 So, I mean, this still leaves 25% of people that are like upset about it, but or maybe it was a smaller number than that. 00:08:46.080 --> 00:08:47.120 I'm I'm not sure exactly. 00:08:47.120 --> 00:09:01.200 But the point is that like we have this process where everyone comes to the meetings, we have all the points of view in the room at the same time, everyone gets to have their say, sometimes for way too long, and then finally we come to a vote, and people are persuaded or they're not persuaded. 00:09:01.200 --> 00:09:09.600 And the idea that we missed something, there was an oopsie, and the design has defects in it that no one realized before is just incorrect. 00:09:09.600 --> 00:09:15.919 Like all of the things that have been said since the CD and since the disc went out have were said before either one of them went out. 00:09:16.240 --> 00:09:16.320 Right. 00:09:16.320 --> 00:09:18.559 There has not been substantial new information. 00:09:18.559 --> 00:09:23.919 And that that is why that's one of the reasons why I point to this as a fundamental breakdown in the process. 00:09:23.919 --> 00:09:32.320 And what's interesting, if you read the comments, which you and I can read, but the general public cannot, because this is not an open and transparent process. 00:09:32.320 --> 00:09:33.200 This is ISO. 00:09:33.360 --> 00:09:33.519 Right. 00:09:33.759 --> 00:09:39.039 Um but if you read the comments, yes, the majority of the comments are about contracts. 00:09:39.039 --> 00:09:47.600 But there are some comments on from people from NBs that voted no that say, yeah, we we think a contract is bad, we should take contracts. 00:09:47.600 --> 00:09:49.519 But also here's some other things that we think about. 00:09:49.840 --> 00:09:51.279 Yeah, reflection was one of the things I heard mentioned. 00:09:51.519 --> 00:09:53.919 Yeah, reflection senders, there's a couple other things. 00:09:53.919 --> 00:09:58.559 And this that that's sort of what I'm referring to. 00:09:58.559 --> 00:10:03.279 That there's we've never had no votes in the past, we've never had this quantity of no votes in the past. 00:10:03.279 --> 00:10:21.840 Clearly, something has broken down in our process that a lot of people felt strongly enough to vote no at this stage, not with a lot of new information, but with information that had already been surfaced, but they were just not convinced that this standard was baked enough, or they felt like they weren't heard enough. 00:10:21.840 --> 00:10:38.799 I've heard you know that there are some people that say that that the policies and norms that the committee has been using for a few years have been sort of suppressing minority voices, and that this is those minority voices making themselves heard. 00:10:38.799 --> 00:10:51.679 There's some people who point to the fact that the stakehol the stakeholder model that actually votes on the standard, which is national bodies, is not the same as what votes to ship the draft out. 00:10:51.679 --> 00:11:05.279 That you know, you could have it be the case where you know 80% of the people in the room vote in favor of a thing, 80 per 80% of the experts, 20% vote against, but those 20% represent 12 national bodies. 00:11:05.279 --> 00:11:22.320 Um but but usually like one part of our process is to not just look at the numbers, but also to look at who like the national body positions before like the convener will usually go and ask the national bodies, you know, you know, we saw that somebody from your NB voted no on this. 00:11:22.320 --> 00:11:23.919 Does this represent a national position? 00:11:23.919 --> 00:11:25.679 Would you vote no on the standard on this? 00:11:25.679 --> 00:11:35.360 And for this to have happened at the scale that it happened really, I think, indicates a breakdown of trust and faith and communication on the committee. 00:11:35.840 --> 00:11:39.759 So I I don't I don't think that I mean, well, I don't get the impression that's what's going on anyway. 00:11:39.759 --> 00:11:53.600 I I don't know if you may be right about some or all of that, but my impression is that what happened was and especially because if you look at how the national bodies voted or the the way that they were producing comments from the CD to the dis, right? 00:11:53.600 --> 00:11:58.799 So the the first draft before we have the comments and after we resolve the comments in the second draft is the DIS, right? 00:11:58.799 --> 00:12:02.720 So between those two, all that happened was consensus got greater, right? 00:12:02.720 --> 00:12:06.960 Like all the comments that were addressed addressed some of those issues that people had. 00:12:06.960 --> 00:12:08.799 And there were some that obviously were not addressed. 00:12:08.799 --> 00:12:11.600 We don't we don't take every issue and just radically change everything. 00:12:11.600 --> 00:12:16.799 Sometimes an issue gets rejected because people say that's out of scope or we're gonna do it later or whatever it is. 00:12:16.799 --> 00:12:30.639 And um and so like if you look at the kind of straw poll kind of votes that we had before that, you will see that like a bunch of the votes changed between those two drafts in a way that doesn't make any sense. 00:12:30.639 --> 00:12:32.799 Like there's no new technical information, but they changed. 00:12:32.799 --> 00:12:46.960 And so what it feels like to me is happening is people that were salty about it, who have a lot of pull in the community, went and lobbied with people that they thought they would get a friendly year with on those national bodies, and they just changed people's minds. 00:12:46.960 --> 00:12:48.480 That's what it feels like happened to me. 00:12:48.480 --> 00:12:59.600 I don't have any proof of that, but that that's what I mean, when you see no technical difference and yet you see the votes change radically, like, well, people's opinions changed, but the
technical details didn't change, so what's going on there? 00:12:59.600 --> 00:13:09.200 And so I I kind of think like if you have the system where all the technical people come into the same room and talk about it, and now we have virtual attendance. 00:13:09.200 --> 00:13:17.919 So especially if you have subject matter expertise or interest in a particular thing, it's very easy to get an invite for just one session or for the whole meeting remotely or whatever. 00:13:17.919 --> 00:13:32.879 And if you come for just the session that you care about or just the day you care about, you know, it if it's in the wrong part of the world, it might cost you a night of sleep, but that's like that level of participation isn't there from people that are then going and saying like it's fine, but then it's not fine later on with the same technical stuff. 00:13:32.879 --> 00:13:36.159 It it it it it smells it smells like something's off in Denmark. 00:13:36.159 --> 00:13:37.600 It's it does not smell right to me. 00:13:37.679 --> 00:13:56.480 Like I I think so, but I I also like we for a long time have assumed that voting things into the draft and voting the draft out were sort of the be-all and end-all, and that the rest of it was procedure that we like were for that was forced upon us by ISO. 00:13:56.480 --> 00:14:02.720 Like back in the day, we used to take plenary vote by NB, not by individual expert. 00:14:02.720 --> 00:14:02.879 Right? 00:14:02.879 --> 00:14:04.240 And that actually was. 00:14:04.240 --> 00:14:11.679 And that actually was it well, maybe it was in play when I started, and I just don't remember. 00:14:11.679 --> 00:14:13.200 But that actually is a good idea. 00:14:13.759 --> 00:14:15.519 That was in in in plenary that was the case. 00:14:15.759 --> 00:14:15.919 Yeah. 00:14:15.919 --> 00:14:16.799 That's actually my own. 00:14:17.039 --> 00:14:20.000 Oh no, actually, it was actually something weird, it was a weird hybrid of that. 00:14:20.000 --> 00:14:23.759 It was NBs get a vote, and then everyone from Insights also gets a vote. 00:14:23.759 --> 00:14:24.879 Which is really weird. 00:14:24.879 --> 00:14:27.679 Like, why does the US Insights is the US organization? 00:14:28.399 --> 00:14:37.200 It's because, and I I have railed on this before, but now have this one beautiful week where I am unemployed and can speak more freely. 00:14:37.200 --> 00:14:41.200 The the stakeholder model is fundamentally broken. 00:14:41.200 --> 00:14:50.399 Neither of those neither of the two stakeholder models, either like a committee of experts, neither a committee of experts nor a committee of 300 experts. 00:14:50.399 --> 00:14:59.600 Neither a committee of 300 experts nor a collection of national bodies is representative of the C ecosystem as a whole. 00:14:59.600 --> 00:15:06.240 Um I like the the cost of what has happened here is not great. 00:15:06.240 --> 00:15:12.480 This is gonna delay the C 26 standard, it's probably gonna be the C 27 standard. 00:15:12.480 --> 00:15:24.879 It's gonna delay it at least four to five months, maybe more, because we're assuming all all assumptions about the timeline now are on the basis that we'll be able to reach some agreement in Brazil. 00:15:24.879 --> 00:15:33.039 And I think the Brazil meeting is gonna be consumed by finger pointing and getting to the bottom of what happened here. 00:15:33.039 --> 00:15:34.559 And I'm I think that's a good thing. 00:15:34.960 --> 00:15:35.679 Well maybe. 00:15:35.679 --> 00:15:38.240 So so there's an there's another way it could go, though, right? 00:15:38.240 --> 00:15:39.600 I I understand what you're saying. 00:15:39.600 --> 00:15:49.679 Like, and a lot of people have this fear that, oh, we're gonna have to go back and yank out contracts, or we're gonna have to go back and like achieve consensus some other way and do this whole long process, and maybe the whole week will be taken up by it and so forth. 00:15:49.679 --> 00:15:51.440 But there's another way it could go, right? 00:15:51.440 --> 00:16:00.559 Which is that we could say at the very beginning of the the first day, let's talk about this as much as we want for today or for the first half of today, and then let's vote. 00:16:00.559 --> 00:16:06.559 And if it doesn't get different consensus than it did before, let's just move on with our lives and pretend this didn't happen. 00:16:06.559 --> 00:16:07.600 Like, who cares, right? 00:16:07.600 --> 00:16:14.000 We'll just we'll just this will be 26, the implementers are already implementing it, and then we'll just make it 29, and we'll vote then. 00:16:14.000 --> 00:16:17.440 And if I mean to not to not actually publish a standard? 00:16:17.440 --> 00:16:18.720 Yeah, we'll just skip one. 00:16:18.879 --> 00:16:34.639 Yeah, I read in the there's a paper that got published that I'll link that was had like a bunch of different authors, and it said one of the options was just like skipping C26, which I thought was like Well, it's not it's not skipping C plus plus 26, it's skipping ISO's C plus plus 26. 00:16:35.039 --> 00:16:35.440 That's the thing. 00:16:35.440 --> 00:16:40.639 So we've we basically already produced a PDF independent of ISO that says what C plus 26 is, and it's already out there. 00:16:40.639 --> 00:16:41.120 Well right. 00:16:41.120 --> 00:16:42.399 So I want to talk about it. 00:16:42.399 --> 00:16:49.679 Stamps a particular PDF that says no, this is the official one, and they get to charge you money if you if you pay for it. 00:16:49.679 --> 00:16:58.799 But like other than then, you know, I don't know who's buying that ISO standard these days, but but like other than the fact that ISO puts a stamp on it, like we don't actually need that. 00:16:58.799 --> 00:17:03.840 And there's there's even an argument to be made that we don't need ISO at all. 00:17:03.840 --> 00:17:04.319 Right. 00:17:04.400 --> 00:17:06.160 That that that that's what I strongly believe. 00:17:06.240 --> 00:17:06.400 Yeah.
00:17:06.400 --> 00:17:13.359 So I'm not advocating for that necessarily, but what I'm saying is like I don't think that it's obviously going to cause a big delay. 00:17:13.359 --> 00:17:15.119 I think there's a question of whether it will. 00:17:15.119 --> 00:17:24.079 But but yeah, it it I think the probable thing is that people are gonna want to talk about it for way too long, and it's gonna take them a whole bunch of of uh that week, if not the whole week. 00:17:24.079 --> 00:17:39.200 But there's also the possibility that we say, okay, well, the people that really want to hash this out, you go off in a room and and you kind of you know work on it until you've got something for everyone else to talk about in plenary, and then everyone else goes off to their other rooms, and maybe we make progress that way, even if it's gonna take a long time. 00:17:39.200 --> 00:17:39.920 So who knows? 00:17:40.160 --> 00:17:43.119 We could we the other option is we could just go to F Dis. 00:17:43.119 --> 00:17:49.680 One of the one of the features of F Dis is on a DIS ballot you can vote yes, no, or abstain. 00:17:49.680 --> 00:17:54.319 On an F DIS, all NBs have to vote and they have to vote either yes or no. 00:17:54.319 --> 00:18:01.279 And so we we would and and the can the the barrier for what's required is a little bit lower. 00:18:01.279 --> 00:18:04.799 So there were a bunch of NBs, including the US, that abstained. 00:18:04.960 --> 00:18:05.279 Yeah. 00:18:05.519 --> 00:18:08.799 So I I want to touch a little bit more on the question of ISO. 00:18:09.119 --> 00:18:18.559 I just want to throw I just want to throw my my acronym that I came up with, ISO I know, which is ISO in name only, inspired by the Rhino Republican in Name Only. 00:18:18.559 --> 00:18:20.079 ISO I know, folks. 00:18:20.079 --> 00:18:24.240 Uh it's uh I'm just throwing it out there in case anybody needs an acronym for whatever we do next. 00:18:24.240 --> 00:18:25.839 Or C does next. 00:18:25.839 --> 00:18:28.240 I shouldn't say we there, because I don't have a dog in this fight. 00:18:28.240 --> 00:18:34.720 I'm just eating my popcorn, listening to this back and forth, being like, oh boy, are the ADSP listeners in for a treat. 00:18:35.680 --> 00:18:38.799 So I for once I can say that I'm unemployed. 00:18:38.799 --> 00:18:48.000 My views do not represent the position of my previous employer or my future employer or my current non-existent employer. 00:18:48.000 --> 00:18:58.160 And I've said this before, and I'll say it louder and more specifically here, that I think that the C committee, I think actually all of the programming languages should exit ISO. 00:18:58.160 --> 00:19:01.119 And yes, it is possible to do so. 00:19:01.119 --> 00:19:03.920 I think it can be negotiated and arranged with ISO. 00:19:03.920 --> 00:19:11.279 It does not mean that we do not continue publishing standards as ISO, because ISO does not want to compete with other standards organizations. 00:19:11.279 --> 00:19:19.200 And because of that, ISO has this fast track submission process by which an external standards body can publish their standard through ISO. 00:19:19.200 --> 00:19:28.319 And I think that that's a route that we can do, but the advantage is that if we exit ISO, we can own our process. 00:19:28.319 --> 00:19:31.279 And there's a few things that are fundamentally broken with the process. 00:19:31.279 --> 00:19:35.440 One, it is not an open and transparent system. 00:19:35.440 --> 00:19:44.480 And a lot of people like l like you know make a point of this that like, you know, the the the papers, the minutes, like there's large parts that are not public. 00:19:44.480 --> 00:19:55.119 And I'm not saying that in the future that we would make everything public, because there is an argument that it does have a chilling effect on on people if every every everything that they say in the meeting is broadcast out to the internet. 00:19:55.119 --> 00:20:12.000 However, what people don't realize is that the current status quo, where we're allowed to publish the papers openly, where we're allowed to have some amount of information openly, every couple of years, ISO tries to take that away from not just us, but all JTC standards. 00:20:12.000 --> 00:20:22.319 And there have been pushes recently that there's the parent body that the C committee is a part of, is called JTC One, which is the joint technical committee for all the tech standards. 00:20:22.319 --> 00:20:27.200 And they have to fight with the rest of ISO to have the levels of openness that we have today. 00:20:27.200 --> 00:20:39.359 And even then, the level of openness that the C committee has, if JTC One knew about that level of openness, they would probably crack down on it because it's technically not really kosher. 00:20:39.359 --> 00:20:44.000 It's just sort of like a status quo that's been necessary for evolving the language. 00:20:44.000 --> 00:20:48.880 But a crackdown could come at some point in time, and and they've tried to in recent years. 00:20:48.880 --> 00:20:54.000 The other thing is that the stakeholder model is just wrong. 00:20:54.000 --> 00:21:07.440 The model that we use, where it's a committee of experts at some points in time, and then it's you know, national bodies at some other points of time is just not reflective of the C ecosystem. 00:21:07.440 --> 00:21:13.200 And no other programming language uses this stakeholder model to make decisions. 00:21:13.200 --> 00:21:18.559 Some of them have, you know, a small committee or panel of decision makers. 00:21:18.559 --> 00:21:22.880 Sometimes it consists of representatives from both the vendors and the community. 00:21:22.880 --> 00:21:29.200 In some cases, it's where there's a voting body, but it's a voting body of like a core developers, as in the case of Python. 00:21:29.200 --> 00:21:30.880 There's a lot of different models out there. 00:21:30.880 --> 00:21:34.079 None of them have this model where
national identity matters. 00:21:34.079 --> 00:21:44.480 And I do not think that your opinion on C should be a factor of what country you live in or what country your company is in. 00:21:44.480 --> 00:21:59.839 And in particular, there have been two events in the past couple of years in which the majority of companies on the committee and the USNB's voice has been significantly suppressed that really troubled me. 00:21:59.839 --> 00:22:05.279 The first is we had a convener election, a convener election in which I did not run for convener. 00:22:05.279 --> 00:22:06.799 I had no horse in this race. 00:22:06.799 --> 00:22:09.920 We had two candidates for convener. 00:22:09.920 --> 00:22:15.039 One candidate was nominated by the U.S., one candidate was not nominated by the U.S. 00:22:15.039 --> 00:22:21.200 Due to an obscure rule, because there was a candidate nominated outside of the U.S. 00:22:21.200 --> 00:22:25.759 by somebody that was not by an NB that was not the U.S., that candidate won by default. 00:22:25.759 --> 00:22:39.599 So instead of having a chance for the national bodies to vote on two candidates and pick which of the candidates was more qualified, we picked a candidate solely because this was the candidate that was not from the U.S. 00:22:39.599 --> 00:22:44.640 And the committee did not have a chance to vote on these two candidates based on their merit. 00:22:44.640 --> 00:22:56.559 The decision was made solely based upon where a particular person lived, or more accurately, on what national body they they chose to join and were able to join. 00:22:56.559 --> 00:23:02.079 And the second thing that happened is with this vote, the within the U.S. 00:23:02.079 --> 00:23:08.799 delegation, which is the largest delegation and represents most of the major tech companies and most of the major vendors. 00:23:08.799 --> 00:23:13.759 So every major vendor is represented on the C committee through the U.S. 00:23:13.759 --> 00:23:20.319 delegation, and every one of the biggest companies in the world is represented by the U.S. 00:23:20.319 --> 00:23:21.039 delegation. 00:23:21.039 --> 00:23:26.960 And also the majority of the members of the committee, the majority of the committee members are from the U.S. 00:23:26.960 --> 00:23:29.599 delegation, and they all share one national body vote. 00:23:29.599 --> 00:23:30.480 The U.S. 00:23:30.480 --> 00:23:34.079 committee voted to approve the standard. 00:23:34.079 --> 00:23:35.200 Now, the U.S. 00:23:35.200 --> 00:23:39.920 committee doesn't actually get to vote on the C standard. 00:23:39.920 --> 00:23:40.559 What the U.S. 00:23:40.559 --> 00:23:47.759 committee gets to do is they get to make a recommendation to a parent organization called the Insights Executive Board. 00:23:47.759 --> 00:23:53.519 And then the Executive Board gets to decide whether to go with that recommendation or not. 00:23:53.519 --> 00:23:57.680 And almost always that parent body has gone with that recommendation. 00:23:57.680 --> 00:24:00.000 But in this case, it did not. 00:24:00.000 --> 00:24:09.279 And they also did not communicate that they did not go with this decision until it was late enough in the process that this could not be overturned. 00:24:09.279 --> 00:24:16.559 So there was a lot of secret meetings and communications in which the majority of the members of the U.S. 00:24:16.559 --> 00:24:19.039 delegation were not aware that the U.S. 00:24:19.039 --> 00:24:21.440 was going to vote to abstain. 00:24:21.440 --> 00:24:23.599 And at one point their U.S. 00:24:23.599 --> 00:24:27.759 was going to vote to disapprove, even though everybody in the U.S. 00:24:27.759 --> 00:24:32.240 committee had voted and thought that our position was going to be to approve the standard. 00:24:32.240 --> 00:24:36.559 And there was a lot of back and forth about this, and some things were rectified. 00:24:36.559 --> 00:24:42.720 But at the end of the day, for a second time, you know, people's voices and votes were taken away. 00:24:42.720 --> 00:24:44.880 So we did not get to vote on a convener. 00:24:44.880 --> 00:24:45.599 The U.S. 00:24:45.599 --> 00:24:50.480 delegation did not have their vote on C26 recorded properly. 00:24:50.480 --> 00:24:54.160 And all of these to me are just symptoms of a fundamentally broken process. 00:24:54.160 --> 00:25:16.799 And so I think that I think that the committee as a whole should figure out how to negotiate an exit and establish a separate, nonprofit foundation, specifically like and Standards Institute for lack of a better term, similar to like the Python Software Foundation or the model that Rust uses, or that any of these other languages, like web standards too, have a separate system. 00:25:16.799 --> 00:25:28.720 Establish our own system with our own policies and procedures that is more open, more transparent, has less arcane rules, and has a stakeholder model that's actually representative and lets us make decisions. 00:25:28.720 --> 00:25:31.759 And I think that if that's not done, then C will die. 00:25:31.759 --> 00:25:38.640 So I would encourage everybody to go lobby your local national body and C committee representatives. 00:25:39.039 --> 00:25:44.640 So there's a couple of other things that are like really specific to ISO that are like problematic. 00:25:44.640 --> 00:25:59.119 Like one of them is if you have someone, you know, we have this code of conduct, and if you break it and you do some truly awful stuff to people at the at one of the committee meetings, there is no process or there is no mechanism by which you can be essentially told like you're not welcome here, right? 00:25:59.119 --> 00:26:00.799 Like it doesn't matter what you do. 00:26:00.799 --> 00:26:06.799 Like because the the national bodies are supposed to handle this kind of thing internally. 00:26:06.799 --> 00:26:10.319 And then if you if you escalate to ISO, they're like, well, that's a national body thing. 00:26:10.319 --> 00:26:23.119 Like
we don't we don't we don't have any control of the individual membership, that always comes from the national bodies, and frequently the national bodies, well, I shouldn't say frequently, the the biggest national body of the US has a bunch of member organizations, and they let them do that internally, and they say the same thing. 00:26:23.119 --> 00:26:24.880 They're like, well, we don't have any control over membership. 00:26:24.880 --> 00:26:31.200 People aren't members of the national body for the United States, they're members of NSIS directly, they're members through their membership organization. 00:26:31.200 --> 00:26:35.839 And so that that's bad for like people feeling comfortable to come to meetings. 00:26:35.839 --> 00:26:42.000 If we want to have, and I I do, I think we should have broader participation from kind of rank and file programmers. 00:26:42.000 --> 00:26:49.759 Uh, and that that really limits that if people are gonna be even occasionally doing bad actions that that are just like given a pass, right? 00:26:49.759 --> 00:26:54.079 Like there's a certain group of people just will not come to to your meetings if that's if that's the case. 00:26:54.079 --> 00:27:04.160 And then there's also this thing of like, you know, we have this remit to make this PDF that specifies what's in the C programming language. 00:27:04.160 --> 00:27:10.559 And what is in the language is restricted to exactly what is in the language proper. 00:27:10.559 --> 00:27:15.759 So something like a package manager cannot be part of the current standard, right? 00:27:15.759 --> 00:27:23.759 We had an effort for a while to have like uh an ecosystem standard as well that would have stuff like that in it, but that that ended up not coming to the same. 00:27:23.759 --> 00:27:25.359 I sorted that effort famously. 00:27:25.359 --> 00:27:34.240 Well, and and I don't I don't actually know any of the details of of why it's it's not happening, but the person who was spearheading it kind of stepped away, and uh like I said, I have no idea why. 00:27:34.240 --> 00:27:38.640 So there may have been like ISO-specific reasons for that as well that I don't know. 00:27:38.640 --> 00:28:02.240 But but the point is that like if you were able to say like, well, you know, a modern programming language should have certain things in it, and a standard library is an important element of it that we have, but there's also other important elements like a package manager or a build m maybe build system or whatever thing, or or even just like the the agreed-on conventions that build systems have to conform to if we don't want to have like uh a build system so the official one for C. 00:28:02.240 --> 00:28:06.160 And we've been doing those kinds of things in an ad hoc way, right? 00:28:06.160 --> 00:28:34.319 Like people come together and they'll write papers for like, you know, a bunch of the CMake folks were were involved in this, I should call them the Kitware folks, and you know, they
were saying, like, well, you know, it would be nice if we could write CMake and and upstream patches to Ninja and Make and so forth that that expect this kind of format, this like particular JSON format for this particular kind of build product, and it'd be nice if all the compilers would generate that build product that same way so that they can be consumed by these other tools, and then they all got on the same page and they did part of that, right? 00:28:34.319 --> 00:28:35.279 And so that's great. 00:28:35.279 --> 00:28:42.960 But the thing is that the kind of scope that people are willing to do that, uh at which people are willing to do that, is is is very limited, right? 00:28:42.960 --> 00:28:54.559 Because you you have this limitation inherent to any effort along those lines, which is how much does it cost an implementation to do it, and how do they justify that as a business expense to their management, right? 00:28:54.559 --> 00:29:07.519 And so if I work at Red Hat and I'm working on GCC, how can I say like, well, I want something like, I don't know, portable BMIs for modules, and there's a good business reason for that, like it's very hard to justify to anybody, right? 00:29:07.519 --> 00:29:10.319 And so we end up having non-portable BMIs. 00:29:10.319 --> 00:29:31.119 But maybe if we had as part of this this notion of how C works, the broader ecosystem, and we could say, like, well, we want BMIs to have the certain guarantees for portability, you know, maybe not across all compilers and across all architectures, but some kind of portability guarantees, and let's see if we can do that, or some kind of build or package management kind of formats that we want. 00:29:31.119 --> 00:29:52.960 And so people, you know, at KitWare and at Microsoft and at at you know like Red Hat and places like that who are who are who is whose job it is to implement these things, they want to know that they're not just like spending 18 sprints in a row on the same feature that ultimately people end up not adopting, and their management is like, never again will I let you run off and and spend all my money this way. 00:29:52.960 --> 00:29:57.279 And so you really need the standard to get behind stuff like that, or a standard. 00:29:57.279 --> 00:30:03.759 And you know, if you weren't under ISO, you wouldn't have this narrowing of what the remit can be, which I think is problematic. 00:30:04.160 --> 00:30:06.240 And also, there's a huge overhead. 00:30:06.240 --> 00:30:09.920 Like we ship on a three-year cycle, the C committee. 00:30:09.920 --> 00:30:31.200 And and a large part of that is because there's a huge procedural overhead to everything we do, to how we we actually turn the standard into the final document and ship it out the door, and it's this arcane process that requires multiple levels of approval, and then it has to be sent off to ISO, and then we have to argue with them about how we formatted our PDF, because they would really just prefer. 00:30:31.200 --> 00:30:33.920 Can you please just send us this thing as a Word document? 00:30:33.920 --> 00:30:37.119 That would really that would really be how we would like to operate. 00:30:37.119 --> 00:30:55.039 Despite the fact that famously back in the like late 1990s, early 2000s, there was a huge drama on ISO about the open office like document format because open document format Microsoft uh may have packed the committee a little bit with experts back in the back in the day. 00:30:55.039 --> 00:30:59.279 Anyways, it's funny that they use that they use word docs for everything. 00:30:59.279 --> 00:31:07.519 But the do we think in this day and age, given the nature of our industry today, that a three-year cycle makes sense? 00:31:07.519 --> 00:31:09.359 And I think the answer is obviously no. 00:31:09.359 --> 00:31:10.799 I think we need to ship faster. 00:31:10.799 --> 00:31:13.359 And people are gonna hear me saying ship faster. 00:31:13.359 --> 00:31:16.240 A certain camp is gonna hear me as saying we should ship more stuff. 00:31:16.240 --> 00:31:19.759 I didn't say we should ship more stuff, I just said we should ship faster. 00:31:19.759 --> 00:31:24.559 I think that we should ship more incrementally and we should be able to rev faster. 00:31:24.559 --> 00:31:28.960 Almost every other language revs faster and is able to get updates out faster. 00:31:28.960 --> 00:31:33.519 And C's inability to do this is one of the things that's holding us back. 00:31:33.519 --> 00:31:39.200 There's a lot of small stuff, small changes and fixes that take like three years to deploy. 00:31:39.200 --> 00:31:50.160 And vendors, understandably, are hesitant to roll out those changes before the final standard gets shipped because they don't know whether they're going to be tweaked. 00:31:50.160 --> 00:31:56.880 And in this m this creates a long delay on when just like fixes get out to users. 00:31:57.279 --> 00:32:03.200 And especially because many shops are like one or two standards behind all the time because of their own internal processes, it it's a very long cycle. 00:32:04.000 --> 00:32:04.160 Yeah. 00:32:04.160 --> 00:32:15.200 And and so like I think to remain a competitive language, to remain an important language, C ⁇ has to be able to evolve with less overhead. 00:32:15.200 --> 00:32:17.440 And that that's not going to happen to RISO. 00:32:17.759 --> 00:32:21.920 And so the the overhead in the current process is more or less fixed at two meetings, right? 00:32:21.920 --> 00:32:26.640 Like we have two meetings every three-year cycle, and if we had a two-year cycle, we would still have two meetings. 00:32:26.640 --> 00:32:29.440 And so it would be a much larger overhead. 00:32:29.440 --> 00:32:32.799 So if we ship more often, we would actually ship less stuff, just inherently. 00:32:33.200 --> 00:32:33.920 Which some people would like. 00:32:33.920 --> 00:32:37.440 But but we what I would really like is you know, Rust ships twice a year.
00:32:37.440 --> 00:32:41.119 If we could ship once a year, that would probably that that would be a big improvement. 00:32:41.119 --> 00:32:41.759 And which is the same. 00:32:41.920 --> 00:32:47.279 Well, that would mean one meeting of with the current system, one meeting of changes and two meetings of bookkeeping, right? 00:32:47.279 --> 00:32:48.880 Like we I don't think we could manage that. 00:32:49.359 --> 00:32:53.200 Yeah, it would be impossible to to to accelerate to to even that cycle. 00:32:53.599 --> 00:32:53.839 Yeah. 00:32:53.839 --> 00:32:56.160 You could do it every 15 months, maybe. 00:32:56.160 --> 00:33:02.000 Right, like every four meetings, you could have half the meetings for for work and half the meetings for for overhead, but that's even that's a lot. 00:33:02.000 --> 00:33:23.680 But I mean, yeah, I mean, I've thought for a long time that maybe because of the overhead, like, and because we have this problem that uh and I'm I'm maybe arguing against myself, you'll you'll see in a second, but like we we have this problem right now that you know you get the three-year cycle, and a lot of times towards the end of that cycle, last two or three or four meetings, people are like, uh I I gotta get my thing in, and and people have objections, like, well, whatever, man. 00:33:23.680 --> 00:33:29.440 I just get my thing in, and they start like really advocating for stuff that's not fully baked sometimes. 00:33:29.440 --> 00:33:29.839 Yep. 00:33:29.839 --> 00:33:31.119 And and it's just human nature. 00:33:31.119 --> 00:33:37.039 I just I'm not, you know, I'm not pointing the finger, I've done it myself, and and I realize that I'm doing it, and you know, I try to mod moderate that. 00:33:37.039 --> 00:33:43.599 But but the thing is that like when you know you're gonna have to wait four years instead of a year, it's a big deal, you know? 00:33:43.599 --> 00:33:44.079 Yeah. 00:33:44.079 --> 00:33:47.519 And so so and that all has to do with that amount of overhead, right? 00:33:47.519 --> 00:33:57.759 And so shortening the cycle does make sense for that problem, and and you know, another thing that might ameliorate that a little bit is just having a longer cycle so you have more time to do stuff. 00:33:57.759 --> 00:34:07.119 And so you end up with that less of like fewer things that get sub suggested are like in the last year if you have a four-year cycle or five-year cycle instead of a three-year cycle. 00:34:07.119 --> 00:34:26.880 But like, yeah, that would be so much more painful to wait five years when you wanted something to to you want some new thing, especially if you had something like Reflection, which, you know, I think the the version of Reflection we got in 26 is great, but I've been playing with the all of Barry's changes for, you know, he's got a he's got a branch of LVM he maintains that has these papers implemented in it. 00:34:26.880 --> 00:34:39.119 And um, I'm getting it's actually funny, I've been I've been working on something that I'm gonna propose using the 29 features, and like as he's writing papers, he's telling me about it, I'm like, well, it'd be great, that would fix this problem that I have. 00:34:39.119 --> 00:34:44.159 And he's like, Well, okay, let me just sit down with Claude for a while and like generate the new changes. 00:34:44.159 --> 00:35:06.079 So he's like literally one day he was like, Oh yeah, well, uh so I the the exact example was I said, you know, I want to be able to detect when I'm in this this, you know, I want to be able to open a constex, a constant valve rather block at the end of a class declaration, and based on what's already declared in the class or defined in the class, the defined members so far, I want to be able to take those and generate some code from that, right? 00:35:06.079 --> 00:35:08.000 And so that's gonna add an extra interface. 00:35:08.000 --> 00:35:12.239 This is for an iterator interface replacement for the CRTP way of doing it before. 00:35:12.239 --> 00:35:22.639 And so I said, well, one of the problems is one of the most common ways that people write operator equals is friend bool operator equals, you know, left right equals default, right? 00:35:22.639 --> 00:35:28.320 And I said, so if someone writes that, like they don't write it in the member form, but in the free function form, I can't see it. 00:35:28.320 --> 00:35:31.360 He's like, oh, well, I can just add a thing that tells you about member functions. 00:35:31.360 --> 00:35:32.480 And I was like, cool. 00:35:32.480 --> 00:35:38.639 And then literally the next day, I just pulled the changes, did a rebuild, and I'm using it. 00:35:38.639 --> 00:35:41.440 So it's it's kind of amazing how fast he's able to do these things. 00:35:41.440 --> 00:35:43.840 Now he's got some automation helping him do it. 00:35:43.840 --> 00:35:49.199 But but like if I had to wait five years for these things, I would be pulling my hair out, you know. 00:35:49.199 --> 00:35:57.840 So you know, I I think a shorter cycle would be nice, and it's definitely impossible with ISO to make it a reasonable thing to do a shorter cycle with all the bookkeeping. 00:35:57.840 --> 00:36:12.159 But I I will say, like, even if we didn't do the ISO cycle, I would still want to have some kind of early review feedback mechanism like we do that takes up those two meetings where we we send it out to people and we say, like, hey, give us detailed notes and we're gonna fix the things that we think are fixable. 00:36:12.159 --> 00:36:12.480 Yeah. 00:36:12.480 --> 00:36:15.039 I think that that's like a really valuable part of the process. 00:36:15.440 --> 00:36:16.480 Most languages do this. 00:36:16.480 --> 00:36:21.039 Most languages have some sort of beta release process. 00:36:21.440 --> 00:36:23.119 Yeah, but the way we do it isn't a beta. 00:36:23.119 --> 00:36:24.480 That's that's that's distinct, right? 00:36:24.480 --> 00:36:28.480 I mean a beta is actually like something is
implemented here, play with it and tell us what you think. 00:36:28.480 --> 00:36:30.800 This is actually just like look at the spec. 00:36:30.800 --> 00:36:33.199 So it's it's harder for people to do this kind of review. 00:36:33.199 --> 00:36:37.199 It would be nice, I think, nicer if we had an actual working prototype that people could play with. 00:36:37.519 --> 00:36:46.880 Yeah, I mean, uh I would imagine yeah, I I would imagine that would be an important part of our process to keep would be would be the the sending back comments. 00:36:46.880 --> 00:36:58.000 And it and in fact, I would one thing we could do if we were making our own process would be whatever the stakeholder model is, if you are a stakeholder and if you want to remain a stakeholder, you need to send back some comment. 00:36:58.000 --> 00:37:01.679 We need to, we need to like know that you've actually reviewed this thing. 00:37:01.679 --> 00:37:22.719 You know, I don't know exactly how how that, you know, it doesn't have to be a hard rule, but like in spirit, if we had some person or organization that was one of the stakeholders and they were just consistently not just voting thumbs up, looks good to me, with no comments, then it's like that you just make it a part of the culture of like that's not what we're looking for. 00:37:22.719 --> 00:37:31.360 We're looking for people who are gonna really actually look over this and make sure that you know we've asked you to give feedback while it's there's still time to make changes. 00:37:31.360 --> 00:37:37.599 We don't want to learn a year down the road that actually, oh, this is broken or we didn't want this, and why didn't you ask us? 00:37:37.920 --> 00:37:38.159 Yeah. 00:37:38.159 --> 00:37:44.559 I think that the worst feedback I've ever gotten when I asked someone to do a technical review of anything is uh no, it looks good. 00:37:44.559 --> 00:37:51.599 I'm like, no, that's it it it either you didn't read it or or you you have just no opinions on technical things. 00:37:51.599 --> 00:37:53.199 Why did I ask you to review this? 00:37:53.920 --> 00:37:57.840 So are you using uh are you using the AI Mup Sarah? 00:37:58.000 --> 00:37:58.400 Wait, wait, wait. 00:37:58.400 --> 00:38:01.679 Should we before we transition to a different topic here? 00:38:01.679 --> 00:38:03.119 Uh we gotta put like a bow. 00:38:03.119 --> 00:38:08.960 You you see, this is why, you know, Bryce is the talker and I'm the the storyline crafter, you know. 00:38:08.960 --> 00:38:11.920 He just like immediately like, oh, AI squirrel, you know? 00:38:11.920 --> 00:38:16.079 Like what is that uh insert clip from up where the dog sees the squirrel. 00:38:17.199 --> 00:38:17.599 Speak! 00:38:17.760 --> 00:38:21.360 Hi there did that dog just say hi there? 00:38:21.599 --> 00:38:22.159 Oh yes. 00:38:22.159 --> 00:38:24.559 My name is Doug. 00:38:24.559 --> 00:38:27.199 I have just met you, and I love you. 00:38:27.199 --> 00:38:29.679 My master made me this
collar. 00:38:29.679 --> 00:38:33.760 He is a good and smart master, and he made me this collar so that I may talk. 00:38:33.760 --> 00:38:34.480 Squirrel! 00:38:35.599 --> 00:38:38.800 Uh what's so what's the what's gonna happen? 00:38:38.800 --> 00:38:41.199 Are we just we're just waiting and seeing, you know? 00:38:41.519 --> 00:38:45.840 So I mean I can I can say I think one of two things is almost certainly gonna happen. 00:38:45.840 --> 00:38:56.719 And those two things are we're going to yank out contracts and then resubmit another CD or DIS or whichever whichever form we have it in. 00:38:56.719 --> 00:39:00.559 So first draft and get a bunch of comments or second draft or whatever. 00:39:00.880 --> 00:39:04.239 Does it have to go to C D or does or can we just send it to NFC? 00:39:04.480 --> 00:39:05.119 That's that's why I don't know. 00:39:05.119 --> 00:39:06.559 I mean because this has never happened before. 00:39:06.559 --> 00:39:10.719 I'm not there's probably a process that I've just I'm not familiar with, so I really don't know what what happens. 00:39:10.719 --> 00:39:16.000 But but I I think like we're gonna send out another version of it without without contracts is a very likely outcome. 00:39:16.000 --> 00:39:27.679 And I think the other out likely outcome is people are gonna vote again in that room, see that it still has the same consensus, and then a bunch of people are gonna object to making changes, and then we're not gonna know what to do at that point. 00:39:27.679 --> 00:39:30.400 And then it's gonna have to be like people at the convener level. 00:39:30.400 --> 00:39:33.519 There's like uh a convener, but there's a bunch of co-conveners for the first time. 00:39:33.519 --> 00:39:38.800 So there's like a group of people that they're making decisions at the top of leadership level these days. 00:39:38.800 --> 00:39:45.199 Probably they're gonna go and have some kind of a discussion about it and and then tell us like what they think we should do to proceed. 00:39:45.199 --> 00:39:46.639 But I I don't know what that would be. 00:39:46.639 --> 00:39:50.320 But I I think that those are the two likely things to happen in terms of what people in that room are gonna do. 00:39:50.559 --> 00:39:51.920 Were the Reddit comments correct? 00:39:51.920 --> 00:40:05.039 Because like I said, I'm very uninformed, so you uh I'm asking the experts here, were they correct though in saying that a bunch of other features and stuff in the standard are built on top of contracts, and so like when you say yank No, only one. 00:40:05.280 --> 00:40:16.239 Only one, so it's not that it's gonna have to But the one can be specified in terms of contracts or not, and we'll just have to go back to there's a previous version of it where it was not specified in terms of contracts, we'll just have to revert back to that. 00:40:16.239 --> 00:40:20.800 It's not a huge it there's work to be done, but it's not like a stumbling block. 00:40:21.199 --> 00:40:35.440 So so to answer the question, so if the disc is either approved or disapproved, but technical changes are required based on the comments, then an F disk becomes mandatory. 00:40:35.440 --> 00:40:42.400 I think that I think that it is correct that C has never had a no vote before. 00:40:42.400 --> 00:41:04.960 However, I do think we've had an F DIS before, and I think that was because we had approved, we had an approved disk ballot, but we had meaningful comments, and that may have been the auto-putter con comments where Ireland had voted, and I think it was Ireland had voted to approve, but it said, hey, this thing kind of looks broken. 00:41:04.960 --> 00:41:11.679 So if you need to make changes to your disc, then you have to go to F Disc. 00:41:11.679 --> 00:41:18.079 And then F Disc has a accelerated voting period. 00:41:18.079 --> 00:41:36.320 Only editorial comments are allowed during the voting period, and the F DIS is a yes or no vote, and you need a two-thirds majority of the P members of the committee, and you need not more than one quarter of the total votes cast to be no's. 00:41:36.320 --> 00:41:49.760 And if that fails, then it just gets sent back to the to the technical committee, and they they go through the whole process again of sending out a diss and sending out a C D. 00:41:49.760 --> 00:41:54.639 So if we took an FDIS and it failed, it would mean a delay of almost a year. 00:41:54.639 --> 00:42:01.199 But we don't have but if we make changes to the document, we don't have to go through another CD and a disk ballot. 00:42:02.000 --> 00:42:02.159 Okay. 00:42:02.719 --> 00:42:15.280 So so if we if we send out an F D, if we'd prepare a draft, we would probably send out a draft sometime by like the end of the year, and then there would probably be an FDIS ballot in the first two months or so of the new year. 00:42:15.280 --> 00:42:27.519 I guess there's some there's some period of time that's needed for so this meeting's November, so we're probably not talking about the F Disc going out until December, and then ICE is probably not gonna submit it until like January or so. 00:42:27.519 --> 00:42:36.079 So we we might have an answer on whether C27 is gonna be published by the first the first meeting of 2027. 00:42:36.079 --> 00:42:57.280 And and if if it's if the F if we can't get an F disk disc to pass, then it really seriously threatens the the continued operations of the committee, because then the committee has to make this decision about do we do we try to publish a standard or do we just keep working on 29? 00:42:57.280 --> 00:43:05.840 And if we say we're gonna try to publish a 27 standard after a failed FDIS, then that's gonna eat at least a year of time for C29. 00:43:05.840 --> 00:43:10.880 And then people are gonna be like, okay, should we push back C29 for another year? 00:43:10.880 --> 00:43:23.519 And if we decide that no, we're not gonna publish the standard, we're just gonna publish the 29 standard,
there's gonna be a whole host of people who are gonna say, no, my industry for regulatory reasons or whatever reason, we need a published ISO standard. 00:43:23.519 --> 00:43:29.840 We can't just have this thing that's an informal, you know, PDF that you threw up on on some website. 00:43:29.840 --> 00:43:31.840 And uh yeah, it's not gonna be great. 00:43:31.840 --> 00:43:34.000 So we really need an F disk to pass. 00:43:34.320 --> 00:43:39.199 Yeah, and because of all that, I mean I think there's a whole lot of people that are like, you know what, I don't care about contracts as much. 00:43:39.199 --> 00:43:44.800 You know, I mean there there's a there's a core of people that really love contracts and are very passionate about it. 00:43:44.800 --> 00:43:56.320 And there's some of the people that I think are really broadly respected on the committee, and you know, one or two that aren't, but but but I mean I think that like the like those people are very passionate about it. 00:43:56.320 --> 00:43:59.440 And I think everyone else is kind of like, yeah, but I mean, like, I don't really need this. 00:43:59.440 --> 00:44:04.639 I Assertions, and that's not the whole story, that's not a very accurate thing, but a lot of people still have that attitude. 00:44:04.639 --> 00:44:07.840 So so I think that that that's why I said that one first. 00:44:07.840 --> 00:44:11.280 That I think the most likely outcome is that we're gonna actually just yank them. 00:44:11.840 --> 00:44:26.559 Honestly, if if we just rebalot it right now, just with an F disc with no changes, we might get an approval by some thin margin solely because solely Well, but nine against is more than twenty-five percent. 00:44:26.559 --> 00:44:28.000 That's the Yeah. 00:44:28.000 --> 00:44:29.920 Yeah, that's you you're right, you're right. 00:44:29.920 --> 00:44:30.960 Against Yeah. 00:44:31.119 --> 00:44:55.440 And so I mean I think that like even if you got two-thirds for and then that many people against, I mean, that's you know, we would consider that to be questionable consensus if there was like a bunch of of strongly against people in the room that were like, you know, it was two-thirds in favor and then a quarter strongly against or something, and then people would be like, okay, look, you all need to go off by yourselves, like the people who wrote this paper and the strongly against people, you need to come to a better consensus and come back. 00:44:55.440 --> 00:44:55.840 Right? 00:44:55.840 --> 00:44:59.920 That's that's a a likely thing that a r someone leading a particular room would say. 00:44:59.920 --> 00:45:07.360 So uh it's not an unreasonable thing, I think, you know, to to try to gain more consensus amongst the national body votes. 00:45:07.360 --> 00:45:22.159 But my fear is that if the way we got to those national body votes was not technical, but rather political, that is to say someone went around and campaigned for this thing, even when there's no real substantive
technical changes, then I just fear that like we're gonna end up doing this over and over again the next few years. 00:45:22.159 --> 00:45:30.239 Every time someone who has a lot of pull with the community doesn't get what they want, they know that they've got this in their back pocket, they can just go around and raise a big stink about something they don't like. 00:45:30.239 --> 00:45:33.440 And you know, you've heard a lot of stuff about profiles. 00:45:33.440 --> 00:45:35.360 I basically think there is no such thing as profiles. 00:45:35.360 --> 00:45:39.599 I've not seen a design that makes any sense that that has been communicated to the committee at all. 00:45:39.599 --> 00:45:44.159 And a lot of the people that are that are involved with that are extremely well known in the committee. 00:45:44.159 --> 00:45:49.119 And if they go around and say, like, you can't ship C without profiles, are we gonna have this again? 00:45:49.119 --> 00:45:56.639 Like, I that's the kind of thing that I'm afraid of, is is that if this turns out to be a viable tactic, then then what is it gonna do to the process? 00:45:56.960 --> 00:46:04.960 Yeah, and I I want to emphasize, and I've said this in a few other places to a few other people, I do not have a horse in this race. 00:46:04.960 --> 00:46:09.440 I do not care about the particular feature. 00:46:09.440 --> 00:46:18.719 And what what I saw, I mean, obviously I I I care to some degree, but I I I'm not invested in this, I'm not even that read up on this particular issue. 00:46:18.719 --> 00:46:28.480 I know people have strong opinions on either side, but for me, the thing that I'm concerned about is the overall health of the language and the overall process. 00:46:28.480 --> 00:46:36.960 And I think it's important to highlight the the failure as a committee separate from the techno. 00:46:36.960 --> 00:46:45.599 That this was this was you know, this was a failure of the system, a failure of of us as a body, that this this this should not have happened. 00:46:45.599 --> 00:46:51.119 Regardless of what outcome you should have wanted, this is not the outcome that you should have wanted. 00:46:51.119 --> 00:46:57.039 If you if you didn't want contracts, you should not have wanted to get contracts taken out this way. 00:46:57.039 --> 00:47:01.519 If you did want contracts, you should not have you obviously would not have wanted this. 00:47:01.519 --> 00:47:11.440 But nobody should have wanted this outcome because this was a bad outcome, and it leads to at best a mediocre outcome, but maybe a much worse outcome. 00:47:11.440 --> 00:47:13.199 And it's a bad look for the language. 00:47:13.199 --> 00:47:14.559 It's a bad look for the community. 00:47:14.559 --> 00:47:16.639 It introduces a lot of uncertainty. 00:47:16.639 --> 00:47:21.840 And so, regardless of your stance on this issue, you should not want it to have happened. 00:47:21.840 --> 00:47:25.280 You should not have wanted the debate to have taken this direction. 00:47:25.599 --> 00:47:26.480 Yeah, yeah, I agree. 00:47:26.800 --> 00:47:36.880 I think the most likely outcome is that an F-DIS passes with some number of remaining no votes below the threshold for consensus. 00:47:36.880 --> 00:47:45.920 I think a slightly better outcome is that we get no no votes at all because we send like uh and sorry, to clarify, I'll make my prediction even more accurate. 00:47:45.920 --> 00:47:50.000 I believe that we will send out an F-DIS draft that has contracts removed. 00:47:50.000 --> 00:47:57.760 I I think that that draft with contracts removed will probably have some no votes from the other side. 00:47:57.760 --> 00:48:08.719 People who are like, well, I'm upset now that you took out contracts and I feel very strongly that contracts was a good feature, so I'm gonna vote no on the standard because you voted no on the standard to take contracts out. 00:48:08.719 --> 00:48:11.440 I'm gonna take vote no on the standard to put it back in. 00:48:11.440 --> 00:48:16.000 So I think that there's that there may be a couple votes like that, but I think that that will likely pass. 00:48:16.000 --> 00:48:20.639 And I think if contracts do get taken out, that's probably the the end of it for contracts. 00:48:20.639 --> 00:48:29.760 I would expect there'll be some some uh some like attempt to put it in a TS, but this is the second time that contracts have been put in. 00:48:29.760 --> 00:48:32.079 Keep in mind it was put in for 20 and then taken out. 00:48:32.079 --> 00:48:33.119 Or was it 23? 00:48:33.119 --> 00:48:33.760 It was 20. 00:48:33.760 --> 00:48:35.519 It was 20, it was putting in for 20 and then taken out. 00:48:35.760 --> 00:48:37.920 It was in for a while and then got yanked. 00:48:38.000 --> 00:48:39.119 Yeah, it was in for a while. 00:48:39.119 --> 00:48:43.199 So if it if it gets taken out, I don't think that it's going back in. 00:48:43.199 --> 00:48:49.920 And uh I think that the most likely thing is that we vote C26 without contracts out sometime in 27. 00:48:49.920 --> 00:48:54.480 And I don't know whether we call that C27, and I don't know whether vendors implement it. 00:48:54.639 --> 00:48:56.719 I think we'll I think we'll still call it 26. 00:48:56.719 --> 00:48:58.880 I I would imagine, but I I I couldn't swear to that. 00:48:58.880 --> 00:49:01.440 But I I think it's very unlikely we're gonna change the train model. 00:49:01.440 --> 00:49:05.280 We're gonna be like, okay, this cost us some time in this cycle, but you know, so did COVID, right? 00:49:05.280 --> 00:49:06.559 So these things happen. 00:49:06.559 --> 00:49:10.000 We're gonna be like, you know, uh 29 is gonna happen because it's the schedule. 00:49:10.079 --> 00:49:13.679 So yeah, I think a lot of it depends on the timeline. 00:49:13.679 --> 00:49:16.639 I think that um yeah, I think a lot of it depends on the timeline. 00:49:16.880 --> 00:49:17.360 Yeah, maybe. 00:49:17.360 --> 00:49:24.079 But I mean I think in terms of like
can I use it in my latest version of GCC, Clang, etc., that's gonna have more to do with what implementers do. 00:49:24.079 --> 00:49:33.679 And I think what they're gonna do is say, like, well, we're just gonna implement the stuff that's in the the published standard that's out there being ratified by ISO. 00:49:33.679 --> 00:49:35.360 They're not gonna wait till ISO's done with it. 00:49:35.360 --> 00:49:37.360 They're just gonna take the PDF that that they get from. 00:49:37.920 --> 00:49:44.719 I don't think that there are any standard, I don't think that there are any vendors loudly on the pro contract side of this. 00:49:44.719 --> 00:49:48.639 So I don't expect I don't think anybody's got an implementation of it that they're shipping. 00:49:48.639 --> 00:49:49.519 Is that correct? 00:49:49.920 --> 00:49:51.679 Oh, I I'm not sure if there's one or not. 00:49:51.679 --> 00:49:53.039 I I don't know uh either way. 00:49:53.039 --> 00:49:53.760 I don't know enough. 00:49:53.760 --> 00:50:02.719 But I know there's other like you know major language and library features that people have already started doing from 26. 00:50:02.719 --> 00:50:16.159 So they're well what that's my point, is that they're on their way, and if if this gets delayed through the ISO process, they're just gonna do everything but the contracts part while they're trying to figure out what it they're gonna leave that one for last essentially because they're not sure about it unless they're sure about it, in which case they'll just keep going. 00:50:16.159 --> 00:50:18.159 In either case, they're gonna keep implementing stuff. 00:50:18.159 --> 00:50:22.079 It's not gonna matter that ISO produced it in 27 or 28 or whenever. 00:50:22.480 --> 00:50:38.719 The the the nightmare scenario would have been imagine if a vendor had implemented contracts and shipped it under a dash std equals C26 flag, and then we took it out, that would have required, I think, a C plus plus 27. 00:50:38.719 --> 00:50:42.079 That vendor would have probably added a std C equals 27. 00:50:42.079 --> 00:50:50.320 Because if they had customers that were starting to use the feature, and that would have really been if that had happened, that would have been very bad for the health of World Overall. 00:50:50.480 --> 00:50:50.719 Yeah. 00:50:50.880 --> 00:50:54.480 Uh one second, girlfriend need the doggy has not eaten yet. 00:50:54.480 --> 00:50:56.320 The doggy napped and it was very cute. 00:50:56.639 --> 00:50:58.559 This is what we call great radio. 00:50:58.559 --> 00:51:02.320 The dog has not eaten yet. 00:51:02.320 --> 00:51:07.360 That needs to be one of the pull quotes that ends up in the lead end to the episode. 00:51:07.360 --> 00:51:08.880 The dog has not eaten yet. 00:51:11.920 --> 00:51:19.280 Ramona's just thrilled that I'm gonna be in the office because Modal is headquartered in New York, so she's not gonna have to deal with me and my loudness. 00:51:19.519 --> 00:51:20.880 Isn't that right, honey? 00:51:20.880 --> 00:51:23.920 It's like that's a lot of people's dream, you know. 00:51:24.719 --> 00:51:25.440 When do we get to the city? 00:51:26.639 --> 00:51:28.079 When do we get to not deal with your loudness? 00:51:28.079 --> 00:51:28.559 Yeah, yeah. 00:51:29.119 --> 00:51:33.760 You gotta leave that in though, so that her episode, so that the number of episodes that Ramona appears in. 00:51:33.760 --> 00:51:34.320 Oh, yeah, yeah. 00:51:35.920 --> 00:51:36.880 This is a great transition. 00:51:36.880 --> 00:51:46.800 Be sure to check these show notes either in your podcast app or at adspthepodcast.com for links to anything we mentioned in today's episode, as well as a link to a GitHub discussion where you can leave thoughts, comments, and questions. 00:51:46.800 --> 00:51:47.599 Thanks for listening. 00:51:47.599 --> 00:51:49.440 We hope you enjoyed and have a great day. 00:51:50.239 --> 00:51:55.360 Low quality, high quality, that is the tagline of our podcast. 00:51:55.599 --> 00:51:56.559 That's not the tagline. 00:51:56.559 --> 00:51:59.599 Our tagline is chaos with sprinkles of information.
Modifica trascrizione