WEBVTT 00:00.000 --> 00:09.640 Cool, I like to do people putting their hands in the air in the talks because it stops 00:09.640 --> 00:15.520 you from falling asleep and it makes me more visibly aware that people are actually listening 00:15.520 --> 00:16.520 to what I'm saying. 00:16.520 --> 00:22.160 So put your hand in the air if you've ever used homebrew, okay, reasonable number of people, 00:22.160 --> 00:27.040 put your hand in the air if you ever contributed to homebrew, put your hand in the 00:27.040 --> 00:36.040 air if you ever tried to contribute to homebrew and I closed your PR, okay, sorry, yeah, 00:36.040 --> 00:42.120 so I'm, we've been, I've not been able to hear what we've talked today, I do follow 00:42.120 --> 00:47.040 try as much as I can, closely, what's going on in Package Management World, shout out 00:47.040 --> 00:51.680 to Andrew's written a lot of very good, comparative blog posts on this stuff this year that 00:51.680 --> 00:52.680 I've enjoyed. 00:52.680 --> 00:55.680 So I'm going to try and talk about today, it's like a bit more high level than I might say 00:55.680 --> 01:00.080 with homebrew stuff about, okay, what we've been doing in homebrew now and why and what 01:00.080 --> 01:03.040 can we maybe learn about it in other Package Management. 01:03.040 --> 01:04.040 So who am I? 01:04.040 --> 01:09.640 Hello, I'm Mike McLeod, if you want to know about me now, my website is the best place, it's 01:09.640 --> 01:13.080 all connected to social media and stuff, I don't read anything on social media anymore, 01:13.080 --> 01:16.720 so email me if you want me to actually see it or personal social media if you want to 01:16.720 --> 01:20.040 talk shit about me behind my back and feel great about it. 01:20.040 --> 01:24.160 My background, I guess, like how I've gone to homebrew and stuff, so I've worked 01:24.160 --> 01:29.720 myself where for almost 20 years, I've been working on homebrew for 17 years this year, 01:29.720 --> 01:35.360 I'm the project leader, which is an elected sort of dictatorship, sort of position thing. 01:35.360 --> 01:40.560 I'm a CTPO in my normal life day job, whatever and that sort of vaguely overlaps with 01:40.560 --> 01:42.920 the homebrew, but not very directly. 01:42.920 --> 01:49.760 So why should anyone care about homebrew, Package Manager is why I have to say, I mean, who 01:49.800 --> 01:54.640 can hear cares about Package Manager is, okay, good, I'm glad that someone does other than 01:54.640 --> 02:01.880 Andrew, because otherwise you'll probably cry, but I guess, in my life, the thing that's 02:01.880 --> 02:06.600 great and terrible that homebrew is that it's somewhat taken for granted, like a lot 02:06.600 --> 02:11.680 of Package Manager is just part of the infrastructure of how the internet works, how 02:11.680 --> 02:13.640 I do my job, whatever, right? 02:13.640 --> 02:18.720 So if you go and run, who's run Brewing Stole in the last week, say, okay, right? 02:18.720 --> 02:22.080 Like, you probably didn't really think very much about it, right? 02:22.080 --> 02:25.480 Because you probably just run something and then hopefully it worked. 02:25.480 --> 02:29.760 But then when it doesn't work, then everyone gets upset. 02:29.760 --> 02:34.840 Like most of the time, it only specific things are broken in specific ways for specific 02:34.840 --> 02:40.960 people, but then when it's all broken for everyone, then everyone gets very concerned. 02:40.960 --> 02:47.360 And the emotional reaction that people tend to have to when these tools break is heavy, 02:47.400 --> 02:52.680 because they're so reliant on these tools, like it's really important that it works. 02:52.680 --> 02:58.680 And in some ways, if I can be a little bit philosophical, it scares people a little bit when 02:58.680 --> 03:05.040 they are forced to actually think how many bits of the internet and tools that they rely 03:05.040 --> 03:09.920 on by all these people at all these places all around the world, that they are completely 03:09.920 --> 03:14.240 unable to do their job, unless all of these things connect together. 03:14.240 --> 03:18.560 And often, the people who are the most involved with open source are the people who are 03:18.560 --> 03:23.600 the most forgiving when hungry breaks and the people who are the most disconnected 03:23.600 --> 03:27.720 from open source are the people who are the most angry and sometimes abusive when 03:27.720 --> 03:29.720 hungry breaks. 03:29.720 --> 03:35.680 It's a bit like your toilet, really, because what it's working, you probably don't spend 03:35.680 --> 03:39.120 a lot of time deeply thinking about your toilet unless, like me, you've recently been 03:39.120 --> 03:42.680 in Japan and you're like, wow, they're toilets are so much better here, why is my 03:42.680 --> 03:44.320 toilet like that? 03:44.320 --> 03:50.080 But then, when you toilet breaks, then we just get shit everywhere, like there's chaos, 03:50.080 --> 03:53.760 no one knows what they're doing, no one knows how to fix it, no one really wants to stick 03:53.760 --> 04:00.480 their hands in there, our friends at the beginning of the toilet, we may be at the end. 04:00.480 --> 04:05.360 So most people get very frustrated when they're too slow, another one, bundler is slow, 04:05.360 --> 04:09.360 like everyone often complained about that for a good many years, right? 04:09.360 --> 04:12.640 And then, what gets done, people gets frustrated. 04:12.640 --> 04:17.880 So in, I feel like in many ecosystems, people get the same feedback about things being 04:17.880 --> 04:22.160 or feeling slow, they maybe don't even know why, but then sometimes the reason why it feels 04:22.160 --> 04:24.640 slow is because we get new tools. 04:24.640 --> 04:29.960 So cargo is often held up as an example of a package manager that feels fast, right? 04:29.960 --> 04:35.240 And I've not done huge amounts of rest up, so I'm not super familiar, but it's, you 04:35.240 --> 04:39.320 know, that reputation for speed is something that's appreciated in the community, it's 04:39.320 --> 04:44.480 sort of found the community's expectations of what these two should be like, how they 04:44.480 --> 04:49.040 should behave, how fast they should feel, how responsive they are, in some ways it kind 04:49.040 --> 04:52.640 of helps to buy into like, what does that language community care about? 04:52.640 --> 04:57.760 Like, the, the Rust community seems to care deeply about building lots of tools that are 04:57.760 --> 05:02.480 responsive and fast, well, paralyzed and all these type of things, you know, a lot of the tools 05:02.480 --> 05:06.800 that, like, one of my favorite tools that's embedded just about everywhere now is like 05:06.800 --> 05:12.320 RIP grip, which, if you were like, how many people went down the path of like, oh, 05:12.320 --> 05:16.880 act, and then the silver structure and finding the perfect tool to search through your code 05:16.880 --> 05:21.600 base increasingly quickly, like, yeah, so if you went down that line, then you learn 05:21.600 --> 05:26.000 a few years ago, like a lot of these two will start to be built in Rust, and they're 05:26.000 --> 05:29.960 just became this little niche of people in the Rust ecosystem who really cared deeply 05:29.960 --> 05:34.560 about just being like, I'm going to make the fastest version of X, or I'm going to make a really 05:34.560 --> 05:38.240 parallelized version of some existing Unix tool that hasn't really been updated in 20 05:38.240 --> 05:39.240 or 30 years. 05:39.240 --> 05:40.240 Right? 05:40.240 --> 05:41.240 UV. 05:41.240 --> 05:42.240 Obviously, that's not an example now, right? 05:42.240 --> 05:48.080 Well, that in the Python ecosystem is kind of reshaped expectations of what speed we 05:48.080 --> 05:51.960 should be expecting from these tools, like, what we can do, and I think UV's an interesting 05:51.960 --> 05:57.760 example, because, again, I'm not super-officer, so excuse me if I miss speaks slightly 05:57.760 --> 06:02.120 on this, but my understanding is some of the reasons why they've been able to be faster 06:02.160 --> 06:06.320 is because they've kind of dropped some support for legacy things, which tools like 06:06.320 --> 06:08.920 people, whatever, might still have to rely on, right? 06:08.920 --> 06:13.040 And, again, like, this is where things get a little bit more tricky, and a bit more 06:13.040 --> 06:18.680 murky for us as, like, package manager, like, developers, it's like, okay, so what, to 06:18.680 --> 06:23.320 what extent do we prioritize speed, or backwards compatibility, or following the 06:23.320 --> 06:27.120 reason why the tools, or maybe we have multiple tools in the same ecosystem, who can 06:27.120 --> 06:28.920 say it, right? 06:28.920 --> 06:29.920 So, bundler, right? 06:29.920 --> 06:34.440 Like, I feel like bundler was a tool that 10 years ago, I heard people complaining about 06:34.440 --> 06:37.960 how slow it was all the time, and then a lot of work went into it, and then now it's 06:37.960 --> 06:42.240 a tool where, you know, it's not the fastest thing in the world, but it's still, like, it 06:42.240 --> 06:48.000 does not seem to get that complete, levied against it, nearly as often as it once did, right? 06:48.000 --> 06:51.880 And I think bundler is a nice example, because they didn't, throughout everything that 06:51.880 --> 06:56.720 they were doing, a lot of just time and effort went into it, behind the scenes, that 06:56.720 --> 07:01.040 meant that the tool ended up just being fast, you can add a few more things, then you 07:01.040 --> 07:04.480 have time send it don't work, then 5% don't work, and if you want to get 100% of the 07:04.480 --> 07:08.720 batches working, 100% of all the time, you essentially have to re-imple that most of 07:08.720 --> 07:12.400 a brew, and that's probably going to be a bit slower than you have right now, right? 07:12.400 --> 07:19.520 So, I ask is all to be collectively both inspired, but cautiously skeptical of some of 07:19.520 --> 07:23.080 these tools that appear out of nowhere, and say, hey, we've been able to do this 10 07:23.080 --> 07:26.840 x-fost and everyone else, because there's some stuff you can learn from that for sure, 07:26.840 --> 07:33.800 like, we have learned from that in Hobroo, but there's also, it's very easy to make a prototype 07:33.800 --> 07:38.120 version of this, which never quite gets around to, if they're writing into something that 07:38.120 --> 07:40.360 works reliably, I can be used at scale. 07:40.360 --> 07:41.360 Thank you. 07:41.360 --> 07:47.360 Why do people do that, not? 07:47.360 --> 07:50.560 In some ways, this reminds me, like, I don't know, how many people will ever sort of like 07:50.560 --> 07:55.080 the legendary Stack Overflow answer that descends into madness about parsing HTML with 07:55.080 --> 07:56.080 rejects? 07:56.080 --> 07:57.080 Yeah. 07:57.080 --> 08:01.760 It reminds me a bit of this, because it's like, you can parse HTML in a simple way 08:01.760 --> 08:05.760 with rejects really quickly and easily, and it's like, look how low code this took, 08:05.760 --> 08:09.960 and it was really quick and efficient, and, like, I'm 90% of the way to solving this problem, 08:09.960 --> 08:13.240 and then I just need to put a bit more time, oh, I'm almost there, I'm almost there, 08:13.240 --> 08:17.680 but then if you know, essentially enough about the problem stays, you know that, like, 08:17.680 --> 08:22.160 fundamentally, you are trying to do something that is, with most of Ajax parses, I know 08:22.160 --> 08:26.480 this pearl does some special things, it's fundamentally impossible, right? 08:26.480 --> 08:31.160 And it's a similar sort of vein here, right, where what they are trying to do is 08:31.160 --> 08:36.080 amrable, but they have taken a shortcut to do so that makes the end result essentially 08:36.080 --> 08:38.440 fundamentally impossible. 08:38.440 --> 08:42.880 So back to home, like, so the, the big thing we did recently in the last year, essentially 08:42.880 --> 08:47.240 was, if you've used home, where you've seen, we do this kind of concurrent downloading 08:47.240 --> 08:48.240 stuff, right? 08:48.240 --> 08:52.320 So, in fact, the person who didn't much of the work, Marcus, is in the room with us right 08:52.320 --> 08:53.320 now. 08:53.320 --> 08:57.480 That's off to Marcus. 08:57.480 --> 09:00.880 And we also have done some work on, like, a slimmer, like, internal API, and stuff 09:00.880 --> 09:06.480 like that, to try and focus on this stuff with speed, when I, as Ruby 4.0.1, like, the 09:06.480 --> 09:10.680 time it takes to out-pahello-wall to a terminal is dramatically slower than the time things 09:10.680 --> 09:12.080 back to do the same thing. 09:12.080 --> 09:17.800 And, okay, most people, like, someone literally with the home-ruh performance stuff, I 09:17.800 --> 09:18.800 personally want social media. 09:18.800 --> 09:23.600 Someone replied saying, in what world this home-ruh performance actually matter to end 09:23.600 --> 09:24.600 users anyway. 09:24.600 --> 09:28.720 Like, yes, this is a good point, but we still get people whining when, like, their shell 09:28.720 --> 09:34.000 initialization takes 0.1 seconds longer than it should have done or whatever, right? 09:34.000 --> 09:38.320 And then, pointing stuff to bash, make things, you know, in our case, sometimes 100 times 09:38.320 --> 09:41.600 faster than just by the time you spin up the Ruby interpreter, right? 09:41.600 --> 09:45.960 So, again, like, it's a weird thing, and it's very project-specific, and it's not 09:45.960 --> 09:46.960 general advice. 09:46.960 --> 09:52.560 But, again, for all of you working-package managers, probably your three items are completely 09:52.560 --> 09:56.760 different, and some naive person from the outside looking in will probably have ideas 09:56.760 --> 09:58.480 of why you're slowing what needs to be fixed. 09:58.480 --> 10:01.840 But you will know what these things are, and you need to make sure you put these things 10:01.840 --> 10:05.560 in front of your team, and in front of contributors you might want to come in, so that 10:05.560 --> 10:09.080 they actually know how they can improve the stuff in a meaningful fashion. 10:09.080 --> 10:14.000 So, as a result of that, like, I think Humbru, with the five point overlice, like, people 10:14.000 --> 10:15.880 are complaining about it less. 10:15.880 --> 10:19.920 I think we've done a bit of a bundle of things, where Humbru's definitely not fast compared 10:19.920 --> 10:23.880 to his competitors, but it certainly feels a bit faster than it was before, and I think 10:23.880 --> 10:29.720 we have now room for understanding and improvement of how we can move forwards. 10:29.960 --> 10:36.160 So, act 2 is, let's think about, like, what we're doing in package manager land, okay, 10:36.160 --> 10:39.120 we talked about, like, user expectations and performance and things like that. 10:39.120 --> 10:47.120 But, I think another big part of software in general, right, like, I've taken over as, like, 10:47.120 --> 10:51.200 a CTO of relatively recently, and one of my big things is, like, how can I make things 10:51.200 --> 10:52.520 go faster? 10:52.520 --> 10:56.640 And one way of making things go faster is performance and profile, like an optimization, 10:56.720 --> 10:58.440 and let's, like, do that. 10:58.440 --> 11:03.440 Another way of making things go faster is thinking about what are we doing today, and what 11:03.440 --> 11:07.000 are the things we can stop doing, and if you can just do fewer things that you get this 11:07.000 --> 11:10.760 combinatorial explosion of, once you have these different options in ways of building 11:10.760 --> 11:15.720 packages and their dependency trees and everything like that, such that it became almost 11:15.720 --> 11:19.400 impossible to debug these issues in a meaningful way and support them and stuff like that. 11:19.400 --> 11:23.360 Because, again, we have limited CIO infrastructure, so we're not able to just go and 11:23.360 --> 11:26.320 say, okay, well, we're going to build everything in every combination every time for 11:26.320 --> 11:29.200 every dependency and figure out if that works. 11:29.200 --> 11:34.000 So, instead, we moved essentially to a model of, like, our binary packages, where instead 11:34.000 --> 11:38.040 of going from room to composite, much of your machine, it's, when we get to a table, there's 11:38.040 --> 11:40.400 or we've been about somewhere else and extracts that, right? 11:40.400 --> 11:44.720 So, that was dramatically faster for users, which dramatically less error for users, 11:44.720 --> 11:48.480 but for power users, that results in an all-spacey being like, hey, there's a bunch of stuff 11:48.480 --> 11:53.560 that you could do with homeboom before that you now can't really do, or at least, to be 11:53.560 --> 11:54.800 more exact. 11:54.800 --> 11:57.960 If you want to do this, you need to go over there and do the work yourselves, because we're 11:57.960 --> 12:01.520 not going to do that work for you, which, as far as a lot of room to do, source users 12:01.520 --> 12:07.000 have concerned, is the same as saying, this is now fundamentally impossible. 12:07.000 --> 12:11.960 But the tricky thing with this is now we're in the situation where, because no, everything's 12:11.960 --> 12:15.440 being built on individual user machines, we need to figure out, like, well, what we're 12:15.440 --> 12:18.520 going to support and what we're going to build and all that type of stuff. 12:18.520 --> 12:22.680 So, it's so, no, we're just like, okay, we build one package and one platform, we pick 12:22.680 --> 12:29.920 a macOS version, we build on X8664, great, then eventually we add the Linux support, then 12:29.920 --> 12:33.760 Apple came along with Apple Silicon, and then we have to do, well, I'm 64 support, and 12:33.760 --> 12:37.560 then we decided, because we hate ourselves, we're going to do, oh, I'm 64 support and Linux 12:37.560 --> 12:41.760 as well, and then I'm sure at some point there will be something else, right? 12:41.760 --> 12:47.920 So, like, yeah, we're at risk, we'll come, I'm sure, you can run homeboom in your 12:47.920 --> 12:49.880 poster, whatever, right? 12:49.880 --> 12:55.560 So, if hands up, any folks in here who support, like software on multiple platforms, or more 12:55.560 --> 13:00.160 platforms than they did 10 years ago, it's a bit of a pain in the ass, right, about what 13:00.160 --> 13:05.160 works for 90% of people, most of the time, right, and then that was their primary focus, 13:05.160 --> 13:06.160 right? 13:06.160 --> 13:08.400 So, in homeboom, we've tried to do the same thing, right? 13:08.400 --> 13:13.000 We tried to focus on, like, what the most of our users actually do, and what that looks 13:13.000 --> 13:17.320 like is getting to a point where you realize, well, we're supporting a bit too much, right? 13:17.320 --> 13:22.120 So, we used to support, in theory, at least, every macOS version, and then a few years 13:22.120 --> 13:25.080 ago, we decided we're going to support three, we're going to support the current one, 13:25.080 --> 13:26.760 and the two preceding ones, right? 13:26.760 --> 13:30.960 This makes people upset sometimes, because they're still using their old Mac, and they 13:30.960 --> 13:34.720 feel like apples abandon them, and now we abandon them, and this is really unfair. 13:34.720 --> 13:38.360 But then even then, even with these three versions, without looks like it's, okay, well, 13:38.360 --> 13:40.080 does that mean we have to support all this? 13:40.080 --> 13:45.640 Or actually all of this, which just gets into madness, right, because every, like, line 13:45.680 --> 13:51.680 here, if you're going to support that, like, we're saying, we expect all of 10,000 packages 13:51.680 --> 13:55.360 that we claim that work on these, we're going to work on these all the time, and then when 13:55.360 --> 13:58.600 it doesn't, it's a bug, we're going to fix it, and it's going to break whatever, and then 13:58.600 --> 14:01.600 this comes back to the, like, well, homeboom is slow, right? 14:01.600 --> 14:05.400 And maybe this isn't slower in terms of the behavior, but homeboom is slower to fix bugs, 14:05.400 --> 14:09.880 there's slower to larger PRs, because we have this matrix of issues that we have to deal 14:09.880 --> 14:12.000 with all of the time, right? 14:12.680 --> 14:14.920 Well, that means is you end up having this limit done, right? 14:14.920 --> 14:17.200 So we say, okay, well, we're not going to support everything all the time. 14:17.200 --> 14:21.440 So we now don't build binary packages for the newest versions of macOS, because we're, like, 14:21.440 --> 14:24.760 well, hey, we don't need to necessarily hyper optimize for that. 14:24.760 --> 14:30.000 Be, we probably get reasonable matrix coverage from the otherwise, and see, like, Apple 14:30.000 --> 14:30.800 is dropping this, right? 14:30.800 --> 14:36.800 So on the 9th September, so on September, 2026, we're going to stop building binary 14:36.800 --> 14:41.560 packages for Intel all together, right, on macOS, still on Linux, because that's the 14:41.560 --> 14:43.440 majority architecture there. 14:43.440 --> 14:45.280 But it's like, well, why are we doing this? 14:45.280 --> 14:47.640 Could we indefinitely build this? 14:47.640 --> 14:50.840 Sure we could, but it's going to make the project slower, and we're at a point where 14:50.840 --> 14:54.880 Apple will have announced their plans to get off this. 14:54.880 --> 15:00.680 So again, multi-billion-dollar corporation, 30 volunteers, right, who receive a small 15:00.680 --> 15:06.000 slide-pend, like, we should not be holding ourselves so higher standard than multi-billion-dollar 15:06.000 --> 15:07.000 corporations, right? 15:07.000 --> 15:10.320 On this, we have to fight harder and harder against that in order to have something that 15:10.320 --> 15:11.320 works for our users. 15:11.320 --> 15:13.880 And eventually we get to a point where we're like, well, this is a security feature 15:13.880 --> 15:14.880 anyway. 15:14.880 --> 15:18.040 No, no, we're not interested in doing this anymore, right? 15:18.040 --> 15:21.520 So we used to just go, okay, we have these features that we turn them off, and then now 15:21.520 --> 15:26.280 we have, like, with a major or minor release, we dedicate, then we disable, then we delete 15:26.280 --> 15:27.280 that code, right? 15:27.280 --> 15:30.800 And again, that, if you don't already do something like this in your project, and you 15:30.800 --> 15:35.400 have, like, a clear cadence, please, it's so satisfying, going, and having a little 15:35.400 --> 15:38.920 bit of that code that goes to your pain in the ass, and knowing, putting little markers 15:38.920 --> 15:43.560 and then being like, in six months, I can just delete all of this mess and not have to 15:43.560 --> 15:44.960 think about this ever again. 15:44.960 --> 15:51.360 Okay, yes, it breaks backwards compatibility, but then, if only we, like, one day, I think 15:51.360 --> 15:56.080 someone will invent something, let's hypothetically call it a version control system, where 15:56.080 --> 16:00.560 if they wanted an old version of some code, they could, in fact, go back in time. 16:01.440 --> 16:06.680 It's a crazy idea, but I think, one day, it might pay off, but, you know, people, again, 16:06.680 --> 16:10.040 it's this thing, it's entitlement, we don't have to deal with that, and we shouldn't 16:10.040 --> 16:11.600 have to deal with that, right? 16:11.600 --> 16:14.360 Rust isn't a nice good example of this, right? 16:14.360 --> 16:18.440 Where they just stay, okay, here's our platform support, here's what we have, we have Tier 16:18.440 --> 16:24.000 1, Tier 1 targets, can be thought of as guarantee to work, and in the spirit of open source, 16:24.000 --> 16:27.440 we almost directly stole this, and we now have our own support, here's our group, where 16:27.520 --> 16:32.600 we say, here, we have our Tier 1 platforms, these are what we expect to work, Tier 2 is 16:32.600 --> 16:37.600 like, maybe this is going to work, Tier 3, it's probably going to be broken, and then, 16:37.600 --> 16:42.880 it's two hundred-sucking, we want you to know, like, hey, we told you to do it this 16:42.880 --> 16:50.040 way, you were doing it a different way, someone said, you are sucking, for those 16:50.040 --> 16:55.680 of the back, I will neither confirm or deny that, but yeah, but it's important, right? 16:55.680 --> 17:01.200 If you can just write this stuff down, you will find people are more likely to accept 17:01.200 --> 17:05.000 a thing which you put in a document or a section on your website, even if you repeat yourself 17:05.000 --> 17:09.960 verbatim in issue clients all day long, they will argue with you until the sun goes home, 17:09.960 --> 17:13.600 but if you say, here's our support document, you can submit a PR with a proposal of why 17:13.600 --> 17:17.200 this should be changed, they will never do that, and they will stop arguing with you. 17:17.200 --> 17:25.640 So, there we go, really, we're going to answer the, I, I thought I had to 17:25.720 --> 17:30.640 slamer here, but I think it's maybe gone, but, no, if I, no, it's coming, right, let's skip 17:30.640 --> 17:38.560 over that for a second, right? So, act 3, sustainability stuff, right? So, what do we think 17:38.560 --> 17:44.800 is the most important thing in open source sustainability, right? Money, put your hands 17:44.800 --> 17:52.320 up if you think it's money, the hacklers are the fun of ruining this delightful cell, I 17:52.400 --> 17:56.600 run a thing to hacklers and the fun, have given them a lot of money to do that. 17:56.600 --> 18:04.640 But we live without any money for many years and I work fine, code, like, let's just 18:04.640 --> 18:08.640 fix all, but we can fix human problems with code, right? We just need to write enough, 18:08.640 --> 18:15.480 no, no, no, okay, yeah, okay, yeah, yeah, we don't like, yeah, humans, okay, the humans 18:15.480 --> 18:20.480 we are what matter, right? Like, if you were a maintainer or an open source project, 18:20.480 --> 18:23.600 no matter how much money you have, no matter how much code you have, no matter how many 18:23.600 --> 18:27.480 users you have, if you decide and you're the so maintainer, you're not going to work 18:27.480 --> 18:32.080 in this anymore, this project is effectively dead, right? Someone can fork and they can 18:32.080 --> 18:36.640 take it on and lead from where you left off, sometimes people do that, a lot of the time 18:36.640 --> 18:41.360 they do not, right? So, what this means is we need to take care of the humans, right? 18:41.360 --> 18:45.360 If you're in a project where you've got multiple maintainers, take care of each other, 18:45.360 --> 18:49.600 if it's just you take care of yourself. I apologize in advance, there's a lot of cell 18:49.600 --> 18:53.360 promotion coming because I've written a bunch about this stuff, it's a double-eye care 18:53.360 --> 18:57.600 and awful lot about, and I'm not going to reiterate all my blog posts, but I am just going 18:57.600 --> 19:01.600 to put them on the screen, so you can go and look them up if you're really that bored. 19:01.600 --> 19:07.600 Right, so the first one, I wrote this thing a while ago, I wrote it before AI stuff, 19:07.600 --> 19:12.400 robot-pedantry human empathy, so this idea of like what we should try and do in our projects 19:12.400 --> 19:17.440 is automate as much as we possibly can, but also not automate the stuff that humans are 19:17.440 --> 19:20.880 actually good at. Like, I remember I wrote this colleague in response to seeing projects 19:20.880 --> 19:26.240 that were like automating their messages of being like, you have made a new PR, good job, 19:26.240 --> 19:32.160 you are valued as a human, and it's like, well, that doesn't really mean a lot to people. 19:38.720 --> 19:46.000 What? Let's have the robots do as much as we can, get them to do, again, pre-AI. 19:46.000 --> 19:50.320 This is mainly talking about CI when I'm talking about this, I love an AI, I'm not going to get it to AI today. 19:51.200 --> 19:57.040 And that's, as humans, right, we can tell people like, good job, thank you, you've kicked us, 19:57.040 --> 20:00.400 I'm looking forward to seeing another PR for me, because you did such a good job in this stuff. 20:00.400 --> 20:05.360 That really makes a difference. There are people who contribute to open source for a long time, 20:05.360 --> 20:09.840 just because of nice things one or two people did. Right, all right, there's a guy here at this 20:09.840 --> 20:14.960 conference, most years, Cornelius Schumacher, he was my mentor on Google Summer of Code 2006, 20:15.040 --> 20:18.800 when I worked on Katie, I can say there's absolutely no way I would have done any work on her 20:18.800 --> 20:23.200 brew, where it not for him being a really good, you know, Summer of Code Mentor, right, there's so many 20:23.200 --> 20:27.440 people throughout this entire conference, it's entire industry, that's stories like that, where this 20:27.440 --> 20:31.920 one person made a difference, I'm probably not sure exactly, right, that person could be you, 20:31.920 --> 20:37.600 by just being nice, saying nice things, encouraging people, right, and I know we have to deal with a lot of 20:37.600 --> 20:43.600 toxicity, and it's hard, you can get a thick skin, I have a very thick skin, but trying to free 20:43.680 --> 20:47.200 yourself out for those moments, where you can connect with someone else, can be like, 20:47.200 --> 20:52.800 good job, well done, it makes a big difference, right, next thing, contribute to funnel, right, 20:53.920 --> 20:59.600 very brief TLDR of this, think about you have a nice funnel on your project, right, a bit like 20:59.600 --> 21:03.440 a sales funnel, any of you know what that is, but basically this idea of like the people who use 21:03.440 --> 21:08.080 your project, some small proportion of them will end up contributing to your project, 21:08.080 --> 21:11.440 and then some small proportion of them will contribute enough that you want to make them 21:11.520 --> 21:17.520 containers, right, but like, if you can get that funnel to be as effective as transitioning 21:17.520 --> 21:20.560 between each stage, that's going to be really good. So one of the things, when we're really 21:20.560 --> 21:25.280 early on, there's so many expectations that like, this is not a project run by us for you, 21:25.280 --> 21:30.480 this is a project run by all of us together, right, whom we had no way of doing automatic version 21:30.480 --> 21:34.640 bumps or anything like that, but it first created max, the greater it was just like, right, 21:34.640 --> 21:38.320 I'm not going to point maintainers for every package, I'm just going to say, if you notice this 21:38.400 --> 21:42.080 is outdated, submit a poor first to fix it, right, that did a really good job of getting a 21:42.080 --> 21:46.560 whole brew, a really, really decent number of contributors for the number of users we had, 21:46.560 --> 21:50.880 and we have a tiny number of maintainers relatively, we have like 30, so that's way better than it 21:50.880 --> 21:57.520 was like 10 years ago. Over to our economics, again, just restaying the stuff about like money, 21:57.520 --> 22:04.000 right, like money is not really what matters, like money helps with some problems, if you want to 22:04.000 --> 22:07.760 come to my Ruby James talk later today, you'll hear me talk about how money creates some other 22:07.760 --> 22:17.680 problems. But what we want to use is the money to help humans, right, so we can help the humans 22:17.680 --> 22:22.800 do the right things, because ultimately I think the scarce resource in open source is maintainer, 22:22.800 --> 22:27.520 not even maintainer time, maintainer motivation, right, and if you have a project where the 22:27.520 --> 22:32.240 environment and the users and the support structure and whatever makes you go, this fucking sucks, 22:32.240 --> 22:39.840 I don't want to do this anymore, then that project will die very soon. Cool, so how do you do that? 22:39.840 --> 22:46.240 Say no to things, say no to most things, most of the time, right, like chances are, unless you hear 22:46.240 --> 22:50.320 a feature request a bunch of times or you strongly disagree, you can say no, that's fine, 22:50.320 --> 22:55.680 you're allowed to do that, because ultimately you owe no one anything, read the licenses of 22:55.680 --> 23:00.640 yourself where it says pretty much, and I'm not a lawyer, so please don't do this. If I fuck 23:00.640 --> 23:04.480 up your shit on purpose, you still can't sue me, because you agreed to the license. 23:05.200 --> 23:10.640 Again, I'm not a lawyer to not do this. So what do we talk about today? 23:11.760 --> 23:15.760 Default to fast, right, like, performance, people care about it, 23:15.760 --> 23:18.960 particularly when other tools have lost in the yours, it's not about idea to think about 23:18.960 --> 23:23.360 performance, even if you're just thinking about performance from the scope of what you do with 23:23.360 --> 23:27.520 the support, right, having a support reality that you can publish and you can use that to 23:27.600 --> 23:31.440 set expectations, and then finally, optimize for the maintainers, because they are the people who 23:31.440 --> 23:38.640 you can project a live. So, one change is enough, right, just think about today, the projects you 23:38.640 --> 23:42.880 work on, what's something you could do to help one of these three things. Makes something faster, 23:43.440 --> 23:47.120 maybe clarify even if it's one sentence and you're reading what you support and what you don't, 23:47.120 --> 23:52.160 or set a boundary for yourself or other maintainers for how you can, like, best to support 23:52.160 --> 23:56.000 yourself as a human. That's all who's going to do one of these three things in the next month. 23:58.400 --> 24:01.360 Some people put up their hand while saying, no, we're just confusing. 24:03.200 --> 24:08.160 If you have questions, I don't know if we've got time for any now, we don't, but semi-anymell, 24:08.160 --> 24:13.440 go my love so you've got to have to contact me, and thank you very much for coming.