WEBVTT 00:00:00.001 --> 00:00:05.380 Hello and welcome to Python Bytes, where we deliver news and headlines directly to your earbuds. 00:00:05.380 --> 00:00:11.120 This is episode 233, recorded May 12, 2021, and I'm Brian Okken. 00:00:11.120 --> 00:00:12.160 I'm Michael Kennedy. 00:00:12.160 --> 00:00:13.980 And I'm Marlene Mangami. 00:00:13.980 --> 00:00:18.240 Well, welcome, Marlene. For people who don't know you, can you introduce who you are? 00:00:18.240 --> 00:00:26.100 So I am a Pythonista, of course, and I am based in Hawaii, Zimbabwe. 00:00:26.100 --> 00:00:29.800 I am also really involved with the Python community. 00:00:30.080 --> 00:00:34.160 So I'm currently the vice chair of the PSA Board of Directors. 00:00:34.160 --> 00:00:41.340 The board, I think, for about coming up on four years now, which is really exciting. 00:00:41.340 --> 00:00:44.620 And it's been really a very cool experience for me. 00:00:44.620 --> 00:00:47.760 I'm also a software engineer. 00:00:47.760 --> 00:00:53.780 I work right now with the Rapids team at NVIDIA and have just been doing software engineering with them. 00:00:53.780 --> 00:00:55.600 I will talk a bit about that later. 00:00:55.600 --> 00:00:59.420 But yeah, I'm trying to think what else. 00:00:59.480 --> 00:01:04.940 I'm also a very avid reader and just like doing other things besides software. 00:01:04.940 --> 00:01:07.740 So yeah, that's pretty much me. 00:01:07.740 --> 00:01:08.180 Cool. 00:01:08.180 --> 00:01:08.980 That's awesome. 00:01:08.980 --> 00:01:10.240 You're doing a bunch of cool stuff. 00:01:10.240 --> 00:01:13.700 So I think Rapids seems like a really neat project to work on as well. 00:01:13.700 --> 00:01:17.120 And of course, the Python community side is great. 00:01:17.120 --> 00:01:18.840 So super happy to have you here. 00:01:18.840 --> 00:01:23.340 Brian, you know, having a good readme is really important to a project, wouldn't you say? 00:01:23.760 --> 00:01:24.960 Yeah, definitely. 00:01:24.960 --> 00:01:31.560 And I, for some reason, I don't know, readmes are not difficult to write, but I freeze up. 00:01:31.560 --> 00:01:32.980 It's a blank page syndrome. 00:01:32.980 --> 00:01:37.260 I think often I've gone through and just like copied from some other project. 00:01:37.260 --> 00:01:38.780 What's in there? 00:01:38.780 --> 00:01:39.100 Readme. 00:01:39.940 --> 00:01:45.700 But I don't think that that's like the best way to go about it, really, because sometimes you forget stuff. 00:01:45.980 --> 00:01:49.400 So this was a, we have a recommendation from Johnny Metz. 00:01:49.400 --> 00:01:52.260 It's a tool called readme.so. 00:01:53.180 --> 00:01:55.820 And this is like totally fun. 00:01:55.820 --> 00:02:00.920 It's just this interactive thing where you get to add stuff. 00:02:00.920 --> 00:02:01.980 So we've got the title. 00:02:01.980 --> 00:02:08.680 You can, there's on the left-hand side, there's a bunch of sections where you can select what you want to go into the readme. 00:02:08.680 --> 00:02:15.160 And then it shows a preview on the right, but you can also see the raw markdown. 00:02:15.160 --> 00:02:17.060 And then in the middle, there's an editor. 00:02:17.060 --> 00:02:19.100 So you can actually just edit the whole thing here. 00:02:19.100 --> 00:02:24.020 But really, I don't know if I really would project title. 00:02:24.020 --> 00:02:30.120 What I'd probably do is go through and, you know, pick out to look at what sort of things I'd want. 00:02:30.120 --> 00:02:36.840 So I'd probably maybe some acknowledgements if I took the, if I got some help from somebody, maybe an API reference. 00:02:36.840 --> 00:02:40.800 If it's got a, if it's a library, how to contribute. 00:02:40.800 --> 00:02:42.020 Oh, badges. 00:02:42.020 --> 00:02:43.200 Definitely want badges. 00:02:43.200 --> 00:02:44.260 Oh, yeah. 00:02:44.260 --> 00:02:48.500 And then, you know, maybe like how to run tests if you want to contribute. 00:02:48.500 --> 00:02:54.360 If there's other cool projects using it, I'd want to use by all these sort of things. 00:02:54.360 --> 00:03:00.880 And then you can, the editor only selects, it only shows you the ones, the one at a time, which is nice. 00:03:00.880 --> 00:03:09.000 But then you've got this, this whole generated, really nice looking readme with tables and everything like built in. 00:03:09.060 --> 00:03:12.380 And you can either just copy it or download it and just run with it. 00:03:12.380 --> 00:03:14.860 I think this is really great. 00:03:14.860 --> 00:03:16.200 I'll probably use this in the future. 00:03:16.720 --> 00:03:17.500 I really love this. 00:03:17.500 --> 00:03:24.700 And I'm surprised about the psychological benefit of just showing the little, the section with the one heading. 00:03:24.700 --> 00:03:28.520 So for like example, acknowledgements, you just have hash, hash acknowledgements. 00:03:28.520 --> 00:03:34.180 And then the few things, even though you're editing the whole readme, it seems so much more like, oh, I'm going to just work on that section. 00:03:34.180 --> 00:03:35.260 It's really cool. 00:03:35.260 --> 00:03:36.020 Marlene, what do you think? 00:03:36.020 --> 00:03:37.560 It's really, really cool. 00:03:37.560 --> 00:03:38.300 I like it. 00:03:38.360 --> 00:03:40.100 I think I'm going to try it out. 00:03:40.100 --> 00:03:45.880 I have put no if it at all in this. 00:03:45.880 --> 00:03:49.320 So I think it's, it's something I need to put more if it into. 00:03:49.320 --> 00:03:51.720 And this looks like a really good way to do that. 00:03:51.720 --> 00:03:52.600 So, yeah. 00:03:52.600 --> 00:03:59.480 I think that it would also be great to just like, if you have an existing readme and you want to add some new sections, you're not quite sure how it should look. 00:03:59.480 --> 00:04:04.840 Using this as a jumping point of just to grab sections of a readme to add to an existing one too. 00:04:04.840 --> 00:04:05.540 This would be great. 00:04:05.540 --> 00:04:06.560 Yeah. 00:04:06.560 --> 00:04:08.120 This is really, really cool. 00:04:08.120 --> 00:04:12.080 How do you, how do you, can you start with a new one? 00:04:12.080 --> 00:04:13.940 Like, can I, well, sorry, let me take it back. 00:04:13.940 --> 00:04:15.420 Can I start with an existing one? 00:04:15.420 --> 00:04:17.720 Can I somehow upload an existing one? 00:04:17.720 --> 00:04:18.800 I don't see. 00:04:18.800 --> 00:04:19.180 I don't think so. 00:04:19.180 --> 00:04:21.380 Wait, I can go to raw. 00:04:21.380 --> 00:04:22.080 Hold on. 00:04:22.080 --> 00:04:24.060 Oh, you can probably just drop it into raw. 00:04:24.060 --> 00:04:24.360 Maybe. 00:04:24.360 --> 00:04:26.160 Yes, you can drop it into the raw. 00:04:26.160 --> 00:04:26.540 That's it. 00:04:26.540 --> 00:04:27.100 Okay, perfect. 00:04:27.100 --> 00:04:29.560 You go to raw, which it doesn't hide the sections. 00:04:29.560 --> 00:04:30.860 It's just pure markdown. 00:04:30.860 --> 00:04:31.880 And then you just throw it in there. 00:04:31.880 --> 00:04:32.080 Okay. 00:04:32.080 --> 00:04:34.020 But you can't edit there. 00:04:34.020 --> 00:04:35.640 No, no, but you can flip it back. 00:04:35.640 --> 00:04:37.200 I think probably once you edit it there. 00:04:37.620 --> 00:04:38.980 I don't think you can edit. 00:04:38.980 --> 00:04:41.600 So you can only edit in the, like the editor part. 00:04:41.600 --> 00:04:42.140 So. 00:04:42.140 --> 00:04:42.720 Yeah. 00:04:42.720 --> 00:04:43.860 It still looks really, really cool. 00:04:43.860 --> 00:04:46.020 I've, I've heard of platform as a service. 00:04:46.020 --> 00:04:48.860 I've heard of infrastructure as a service. 00:04:49.080 --> 00:04:52.400 I've heard of database as a service, but I guess now we have read me as a service. 00:04:52.400 --> 00:04:52.720 I don't know. 00:04:52.720 --> 00:04:53.520 You just go to the website. 00:04:53.520 --> 00:04:55.700 Exactly. 00:04:55.700 --> 00:04:56.160 That's pretty cool. 00:04:56.160 --> 00:04:57.420 Yeah. 00:04:57.420 --> 00:04:59.040 I'm, I'm, I'm pretty excited about this. 00:04:59.040 --> 00:05:02.660 Actually, I might, I might play around with this, for my next project. 00:05:02.780 --> 00:05:07.120 I've got some stuff that may end up on PyPI soon and it'd be cool to do it. 00:05:07.120 --> 00:05:07.640 All right. 00:05:07.640 --> 00:05:13.000 So I've got the next item and it's a bit of, a skateboarding dog type of thing. 00:05:13.000 --> 00:05:16.660 It's not something I think a lot of us will take advantage of, but it's something that is 00:05:16.660 --> 00:05:22.580 pretty interesting as we kind of look at how Python is finding its way into the larger 00:05:22.580 --> 00:05:23.500 computing space. 00:05:23.640 --> 00:05:24.080 Yeah. 00:05:24.080 --> 00:05:28.040 And Oh, Sam Morley out there and the live stream before we move on says, it'd be 00:05:28.040 --> 00:05:32.060 really cool if you could point this at a GitHub repo and edit your repo directly. 00:05:32.060 --> 00:05:34.000 the read me directly on your repo. 00:05:34.000 --> 00:05:35.640 Yes, absolutely. 00:05:35.640 --> 00:05:36.380 That's fantastic. 00:05:36.380 --> 00:05:37.860 Yeah. 00:05:37.860 --> 00:05:38.860 That's a really good idea. 00:05:38.860 --> 00:05:39.520 Really good idea. 00:05:39.520 --> 00:05:40.160 All right. 00:05:40.160 --> 00:05:41.480 Back to my skateboarding dog. 00:05:41.480 --> 00:05:48.780 So there's a company called Sarah brass, and this was sent over to, to us by Galen 00:05:48.780 --> 00:05:53.040 Swint, who is a PhD researcher who does high performance computing. 00:05:53.040 --> 00:05:53.880 And stuff. 00:05:53.880 --> 00:05:57.460 So in that world, I think this, this may be a real thing. 00:05:57.460 --> 00:06:01.260 You look through the article here that talks about this announcement and it's like, well, 00:06:01.260 --> 00:06:05.900 there's like these 12 customers or 15 customers of this, this chip. 00:06:05.900 --> 00:06:10.100 But for those of you watching, there's, or you check out the article, there's a woman holding 00:06:10.100 --> 00:06:11.000 a chip. 00:06:11.000 --> 00:06:13.740 And normally we think of computer chips as little tiny things. 00:06:13.740 --> 00:06:20.360 This is a 12 inch by 12 inch computer chip, or you want to go metric 30 centimeters by 30 00:06:20.360 --> 00:06:20.720 centimeters. 00:06:20.920 --> 00:06:23.700 It is a big, big computer chip. 00:06:23.700 --> 00:06:29.340 And the idea is we've had small little chips come along to do special types of processing. 00:06:29.340 --> 00:06:35.960 We've had GPUs come along and do, be adapted, I guess, for things like machine learning, training 00:06:35.960 --> 00:06:36.960 machine learning models. 00:06:36.960 --> 00:06:41.460 This thing just takes that idea to an entire new level. 00:06:41.460 --> 00:06:48.200 So for example, I'm always going on and on and raving about my Mac mini, my M1, where, 00:06:48.200 --> 00:06:51.240 it's a cheap little computer relative to Apple stuff, I guess. 00:06:51.240 --> 00:06:57.120 but it's, it's super fast, but it has four performance cores and four efficiency cores. 00:06:57.300 --> 00:06:57.920 That's it. 00:06:57.920 --> 00:07:01.600 your GPU, if you've got a really high end one might have 4,000 cores. 00:07:01.600 --> 00:07:08.140 This insane little chip here has 850,000 AI cores on one chip. 00:07:08.140 --> 00:07:09.100 Is that insane? 00:07:09.100 --> 00:07:10.380 What do you, what do you do think? 00:07:10.380 --> 00:07:16.000 I, I'm, I'm curious how they, I mean, this is, this is some major advances in, in wafer technology, 00:07:16.000 --> 00:07:19.560 because how do you get that big of a chip with no defects in it? 00:07:19.560 --> 00:07:20.040 Yeah. 00:07:20.040 --> 00:07:23.180 And they have apparently 100%, efficiency. 00:07:23.180 --> 00:07:28.640 Well, first of all, one of the ways you do it is you use the TSMC foundry, who seems to 00:07:28.640 --> 00:07:32.160 be taking over all these small high efficiency, type of things. 00:07:32.160 --> 00:07:36.640 And so they had a previous one that they've more than doubled the core count for. 00:07:36.640 --> 00:07:41.960 And another way to kind of appreciate like how much is going on in this chip, I, you know, 00:07:41.980 --> 00:07:47.880 go back to my, my M1, it has a zero, zero, one, six brilliant transistors. 00:07:47.880 --> 00:07:49.420 This has 2.6 trillion. 00:07:49.420 --> 00:07:50.720 Or is that another way? 00:07:50.720 --> 00:07:55.440 2,600 billion, a billion transistors versus 1.6 billion. 00:07:55.440 --> 00:07:58.660 Like it's a 2,000 times more on this chip. 00:07:58.660 --> 00:08:00.040 So super, super cool. 00:08:00.040 --> 00:08:03.000 And now you may be wondering, all right, all this is interesting and tips are neat. 00:08:03.000 --> 00:08:05.300 What is the Python angle? 00:08:05.300 --> 00:08:08.040 Like, why would I bother putting this on here? 00:08:08.040 --> 00:08:11.280 Cause you know, we don't really talk about chips that much, except for when I go on and 00:08:11.280 --> 00:08:12.000 on about my M1. 00:08:12.000 --> 00:08:12.740 Here's the deal. 00:08:12.740 --> 00:08:17.840 If you scroll down in this article a little bit, you'll see user program, this insane machine 00:08:17.840 --> 00:08:23.860 transparently in machine learning frameworks, such as specifically TensorFlow and PyTorch. 00:08:23.860 --> 00:08:25.940 Isn't that crazy? 00:08:25.940 --> 00:08:27.260 That's really interesting. 00:08:27.260 --> 00:08:28.600 Isn't it? 00:08:28.600 --> 00:08:28.760 Yeah. 00:08:28.760 --> 00:08:31.820 I said, I was just thinking about you as I'm going through this, cause you're working on the 00:08:31.820 --> 00:08:36.040 Rapids project, which is, it's not the same thing obviously, but it's kind of in that space, 00:08:36.040 --> 00:08:36.300 right? 00:08:36.300 --> 00:08:38.120 Yeah, it is. 00:08:38.120 --> 00:08:39.440 Have you heard of this before? 00:08:39.440 --> 00:08:41.260 No, I haven't heard of it. 00:08:41.260 --> 00:08:45.260 This is, yeah, this is really big and I have not heard of it. 00:08:45.260 --> 00:08:46.220 Yeah. 00:08:46.220 --> 00:08:49.260 I will get into reading a bit more about it after. 00:08:49.260 --> 00:08:49.940 Yeah. 00:08:49.940 --> 00:08:51.240 Yeah, for sure. 00:08:51.240 --> 00:08:52.680 So there's a lot of interesting things. 00:08:52.680 --> 00:08:57.180 And one of the, I can't remember where exactly they spoke about it, but they basically say, 00:08:57.280 --> 00:09:00.960 what you do is you program and TensorFlow and PyTorch as normal. 00:09:00.960 --> 00:09:07.740 And then they have this custom compiler that, that rewrites, that extracts this execution 00:09:07.740 --> 00:09:14.660 graph to actually scale out to the 850,000 cores to the developers don't have to think about 00:09:14.660 --> 00:09:16.820 how they program against something like this. 00:09:17.280 --> 00:09:20.460 I don't want to spend too much time on this because there's something, my next item is 00:09:20.460 --> 00:09:20.920 super amazing. 00:09:20.920 --> 00:09:22.440 I want to take the time to dive into it. 00:09:22.440 --> 00:09:25.580 But there's another thing that's really interesting. 00:09:25.580 --> 00:09:30.040 Just as you look at it, like this thing takes an insane amount of power. 00:09:30.040 --> 00:09:36.240 Like, oh, for this one chip, you're going to need a four kilowatt power supply with up to 00:09:36.240 --> 00:09:39.040 a peak power of 23 kilowatts. 00:09:39.040 --> 00:09:40.380 Oh, wow. 00:09:41.180 --> 00:09:45.780 When you plug in an electric car at one of the high speed home chargers, that's seven 00:09:45.780 --> 00:09:47.300 kilowatts, just to give you a sense. 00:09:47.300 --> 00:09:51.100 This is like insane amounts for one chip, right? 00:09:51.100 --> 00:09:52.120 You could think of it as a supercomputer. 00:09:52.120 --> 00:09:53.300 Like it's one chip. 00:09:53.300 --> 00:09:54.020 So anyway. 00:09:54.020 --> 00:09:58.020 Our entire lab doesn't draw that much. 00:09:58.020 --> 00:10:03.140 So the reason I said it's a skateboarding dog thing is I don't think most of us will be 00:10:03.140 --> 00:10:06.140 able to ever even interact with one of these, much less buy one. 00:10:06.140 --> 00:10:09.780 They're going to be shipping in the later part of this year. 00:10:09.780 --> 00:10:13.320 And the price is something like $3 million plus. 00:10:13.320 --> 00:10:15.700 So this is certainly super computer level. 00:10:15.700 --> 00:10:21.500 But I do think it opens the door for really interesting stuff going on in the high performance 00:10:21.500 --> 00:10:22.500 Python space. 00:10:22.500 --> 00:10:23.240 So yeah. 00:10:23.240 --> 00:10:25.040 Glad that Galen sent it over. 00:10:25.040 --> 00:10:29.940 Well, I'm totally going to put $25 into Dogecoin so that I can afford this later this year. 00:10:29.940 --> 00:10:31.420 Oh, speaking of which. 00:10:31.420 --> 00:10:33.120 Exactly. 00:10:33.120 --> 00:10:38.300 Well, what about, I think maybe you get this and you create an AI that can more intelligently 00:10:38.300 --> 00:10:40.560 mine Dogecoin and then you take over the world. 00:10:40.560 --> 00:10:42.500 Just an investment. 00:10:42.500 --> 00:10:43.300 Yeah. 00:10:43.300 --> 00:10:43.960 Just an investment. 00:10:43.960 --> 00:10:45.080 All right. 00:10:45.080 --> 00:10:49.420 So speaking of large scale, high performance computing, Marlene, take it away. 00:10:49.420 --> 00:10:50.160 Sure. 00:10:50.160 --> 00:10:52.720 I have the next item, which is Rapid. 00:10:53.340 --> 00:10:58.020 And I wanted to speak about this because I'm working on it. 00:10:58.020 --> 00:11:00.700 And it's what I've been working on for, I think. 00:11:00.700 --> 00:11:01.140 Yeah. 00:11:01.140 --> 00:11:01.460 Wow. 00:11:01.460 --> 00:11:06.120 It's been about a year since I have been with NVIDIA. 00:11:07.300 --> 00:11:12.180 working as a software engineer there and working specifically on the Rapids project. 00:11:12.180 --> 00:11:19.820 And so Rapids, I think, is really interesting because the goal of Rapids, similarly, like the 00:11:19.820 --> 00:11:23.600 last thing Michael just showed us, is to speed up data science. 00:11:23.600 --> 00:11:25.080 But this is what GPUs. 00:11:25.080 --> 00:11:32.860 So I think it's really, it's been really cool to work on the Rapids project. 00:11:32.860 --> 00:11:37.120 And I think it's really interesting as well because it's open source. 00:11:37.120 --> 00:11:40.180 It's also, there's a lot of Python involved. 00:11:40.180 --> 00:11:44.820 So it's, well, it's not entirely, it's not mostly Python. 00:11:44.820 --> 00:11:49.020 Actually, there's a lot of C++ and Cuda code in there as well. 00:11:49.380 --> 00:11:55.300 But I am not, you know, personally, I'm not, I'm not, my aim is not to learn Cuda. 00:11:55.300 --> 00:11:57.840 It's to try and avoid that as much as possible. 00:11:57.840 --> 00:12:01.440 And also avoid as much C++ as possible. 00:12:01.440 --> 00:12:03.640 So that's a bit more reasonable. 00:12:03.640 --> 00:12:11.740 But one of the goals of the Rapids project is to allow people who are Pythonistas to work 00:12:11.740 --> 00:12:19.140 primarily with GPUs and to get those speed ups without having to know any Cuda code or 00:12:19.140 --> 00:12:21.760 to know any C++. 00:12:21.760 --> 00:12:29.600 And so I have been working primarily on the Python side of things and have really been enjoying it. 00:12:29.600 --> 00:12:34.880 I work specifically on, with the QDF data frame library. 00:12:34.880 --> 00:12:42.920 And QDF is basically a, it mirrors, it's a, it's a GPU data frame library that mirrors pandas. 00:12:43.060 --> 00:12:49.520 So if you have a data set and you'd like to do computations on your data set or do different 00:12:49.520 --> 00:12:55.720 operations on your data set, if you can do that with pandas, you should be able, hopefully, 00:12:55.720 --> 00:12:57.680 to do the same thing with QDF. 00:12:57.680 --> 00:13:03.080 But the good thing is that it will be, it will, it will probably be faster. 00:13:03.080 --> 00:13:10.600 I actually can't definitively say that it will be faster because I remember when I first joined 00:13:10.600 --> 00:13:17.300 the project as well, I was, I really, I'm very enthusiastic and I really enjoy sort of sharing 00:13:17.300 --> 00:13:18.980 when I'm learning something new. 00:13:19.140 --> 00:13:25.300 And I remember I was like going around and speaking and saying that, you know, QDF is 00:13:25.300 --> 00:13:29.000 so much better than pandas because it's just so much faster. 00:13:29.000 --> 00:13:35.740 And then my manager was just like, you need to stop saying that because it's not true all 00:13:35.740 --> 00:13:36.480 of the time. 00:13:36.560 --> 00:13:38.560 it's true most like some of the time. 00:13:38.560 --> 00:13:46.640 So for smaller data sets, it's probably better to stick with pandas because. 00:13:46.640 --> 00:13:47.620 Yeah. 00:13:47.620 --> 00:13:48.980 There's always this overhead, right? 00:13:48.980 --> 00:13:53.960 Like as you scale things out and stuff, there's probably like, well, how do we convert this over 00:13:53.960 --> 00:13:55.240 and get it onto the GPU? 00:13:55.240 --> 00:14:00.440 And if, if that process takes half the time of what just doing the computation, you might 00:14:00.440 --> 00:14:01.660 as well just do the computation, right? 00:14:01.900 --> 00:14:02.260 Exactly. 00:14:02.260 --> 00:14:05.980 Like if you're already, if you're mainly doing, right, I agree. 00:14:05.980 --> 00:14:11.060 Like if you're working with smaller data sets and you are fine with that and that works for 00:14:11.060 --> 00:14:16.640 you and your time is not like being wasted a lot, then I would say, please go ahead and 00:14:16.640 --> 00:14:17.560 stick with pandas. 00:14:17.560 --> 00:14:24.100 But if you have, if you are working on like larger data sets and the larger your data sets 00:14:24.100 --> 00:14:27.600 get, the more the difference is going to be in terms of your speed up. 00:14:27.600 --> 00:14:33.820 So with very large data sets, QDF is going to take a much shorter time to do computations 00:14:33.820 --> 00:14:34.600 and things like that. 00:14:34.600 --> 00:14:35.120 Yeah. 00:14:35.120 --> 00:14:38.820 You've actually put a really interesting example in the show notes here, right? 00:14:38.820 --> 00:14:40.960 Showing how many zeros is that? 00:14:40.960 --> 00:14:44.340 A hundred million items or something like that? 00:14:44.340 --> 00:14:44.820 Yeah. 00:14:44.820 --> 00:14:45.640 It's a hundred million. 00:14:45.640 --> 00:14:51.460 I just kind of like randomly chose a number to try and make it like, I didn't also want to 00:14:51.460 --> 00:14:58.120 take a number that was too big because I didn't want to spend like a long time doing it. 00:14:58.120 --> 00:15:02.600 And I know like for a lot of data scientists, like I think increasingly people are working 00:15:02.600 --> 00:15:06.560 with larger and larger data sets, just depending which field you're in. 00:15:06.560 --> 00:15:10.160 For the example, I put it on the show notes and it's on the screen right now. 00:15:10.680 --> 00:15:16.300 But if you take a Pandas sort of data frame and try and calculate the mean and you take 00:15:16.300 --> 00:15:22.480 the same QDF data set and try and calculate the mean, it will take, I think I'm trying to 00:15:22.480 --> 00:15:22.900 look at the notes. 00:15:22.900 --> 00:15:32.400 It's 105 milliseconds for Pandas and it's like 1.83 milliseconds for QDF, which is just, 00:15:32.400 --> 00:15:33.480 That's awesome. 00:15:33.480 --> 00:15:37.900 And that's like a smaller scale, I would say, data sets compared to some people. 00:15:39.060 --> 00:15:40.660 Yeah, just a hundred million. 00:15:40.660 --> 00:15:42.760 It's just a hundred million. 00:15:42.760 --> 00:15:44.260 So it's not a lot. 00:15:44.260 --> 00:15:45.560 I mean, it depends. 00:15:45.560 --> 00:15:51.400 But yeah, I think it's definitely significant once you get to a certain threshold, which is 00:15:51.400 --> 00:15:51.820 pretty cool. 00:15:51.820 --> 00:15:52.360 Yeah. 00:15:52.360 --> 00:15:52.640 Yeah. 00:15:52.640 --> 00:15:55.360 Over on the rapid site, it's a rapid site AI. 00:15:55.360 --> 00:15:58.340 It says it scales out on multiple GPUs. 00:15:58.340 --> 00:16:04.880 So seamlessly scale from a GPU workstation to multi GPU servers and multi node clusters working 00:16:04.880 --> 00:16:05.840 with Dask as well. 00:16:05.840 --> 00:16:09.760 So Dask is also kind of about scaling Pandas and combining those. 00:16:09.760 --> 00:16:10.700 That's pretty awesome. 00:16:10.700 --> 00:16:14.720 So I actually saw that you have a Dask course out. 00:16:14.720 --> 00:16:19.480 Like I recently thought, yes, definitely going to take that because I mean, yeah, check it 00:16:19.480 --> 00:16:19.580 out. 00:16:19.580 --> 00:16:21.440 That one's awesome. 00:16:21.440 --> 00:16:22.560 Yeah. 00:16:22.560 --> 00:16:22.860 Yeah. 00:16:22.860 --> 00:16:27.660 We put that together with Matthew Rocklin and team over at Coiled. 00:16:27.660 --> 00:16:27.960 Yeah. 00:16:27.960 --> 00:16:28.820 And that's actually free. 00:16:28.820 --> 00:16:30.660 So people can just drop in and take that course. 00:16:30.660 --> 00:16:32.820 I think maybe I can put it in the show notes at the end. 00:16:32.820 --> 00:16:34.620 I think it just just announced. 00:16:34.620 --> 00:16:35.180 Let's see. 00:16:35.180 --> 00:16:36.000 Very cool. 00:16:36.000 --> 00:16:37.060 No, that was a lot. 00:16:37.060 --> 00:16:37.660 That was last week. 00:16:37.660 --> 00:16:38.920 But yeah, this is super cool. 00:16:38.980 --> 00:16:43.420 And this one is certainly within normal person's reach. 00:16:43.420 --> 00:16:45.300 You get a GPU and you're good to go. 00:16:45.300 --> 00:16:45.540 Right? 00:16:45.540 --> 00:16:46.440 Yeah. 00:16:46.440 --> 00:16:46.960 I think. 00:16:46.960 --> 00:16:47.420 Yeah. 00:16:47.420 --> 00:16:50.220 I mean, I'm just using it on my laptop with the GPU. 00:16:50.220 --> 00:16:52.040 You can also use it like online. 00:16:52.180 --> 00:16:58.580 So there's also a CoLab notebook on the rapid side. 00:16:58.580 --> 00:17:02.580 I think that you can click and then you can like kind of experiment if you just wanted to 00:17:02.580 --> 00:17:03.240 do it online. 00:17:03.240 --> 00:17:08.460 Or I think you can use any sort of online GPU that you have access to. 00:17:08.460 --> 00:17:14.280 So it's very, I think it's trying to make it more accessible, which is great. 00:17:14.280 --> 00:17:14.800 Yeah. 00:17:14.800 --> 00:17:15.740 That's super cool. 00:17:15.740 --> 00:17:16.080 Yeah. 00:17:16.080 --> 00:17:16.580 Very neat. 00:17:16.580 --> 00:17:19.400 Well, like I said, I think this is a cool project to be working on. 00:17:19.400 --> 00:17:20.860 So thanks for sharing it with us. 00:17:20.860 --> 00:17:21.400 No problem. 00:17:21.400 --> 00:17:23.980 Ryan, is it time for the next one? 00:17:23.980 --> 00:17:26.000 Is it time for the next one? 00:17:26.000 --> 00:17:27.480 yes. 00:17:27.480 --> 00:17:32.060 This was a recommended by a listener, Ira Horeca, I think. 00:17:32.060 --> 00:17:35.040 So mentioned this in, it's kind of a rabbit hole. 00:17:35.040 --> 00:17:37.580 I spent a whole bunch of time playing with all this stuff last night. 00:17:37.580 --> 00:17:39.320 He recommended Date Finder. 00:17:39.320 --> 00:17:43.680 So this is a Python utility and it kind of is amazing. 00:17:43.680 --> 00:17:45.620 So it's a combination of a couple of things. 00:17:45.620 --> 00:17:53.780 But so he pointed us to a comcode video, which, you know, I'm totally a fan of comcode stuff 00:17:53.780 --> 00:17:57.900 because they kind of go through some of the Python libraries and some of the other, a lot 00:17:57.900 --> 00:18:01.660 of other things, but just have kind of a quick demo of what it does. 00:18:01.780 --> 00:18:03.120 And I really appreciate that. 00:18:03.120 --> 00:18:08.580 It actually, the demo here is better than the read me in the, date finder, read me. 00:18:08.580 --> 00:18:12.180 So, so maybe I guess a pull request is necessary, but anyway. 00:18:12.360 --> 00:18:17.840 So what is, what date finder does is it takes, I'm going to scroll down a little bit. 00:18:17.840 --> 00:18:23.240 So, date finder takes, it parses dates or finds them. 00:18:23.240 --> 00:18:29.520 So it, you give it a string or a bunch of, list of strings or something, and it can find, 00:18:29.520 --> 00:18:31.140 find where the dates are in there. 00:18:31.140 --> 00:18:36.620 So if you've got, if you've got a sentence or a paragraph or, an entire page that has 00:18:36.620 --> 00:18:42.360 a whole bunch of dates in it, it'll find all of them and then return you a list of, 00:18:42.360 --> 00:18:43.680 of dates that it found. 00:18:43.680 --> 00:18:48.080 It actually does a whole bunch of things, but that's the default or the one that 00:18:48.080 --> 00:18:49.320 we're talking about find dates. 00:18:49.320 --> 00:18:54.940 There's a bunch of other, Leslie, less documented, features of, date finder, 00:18:54.940 --> 00:18:58.120 but this, this is the one that is demonstrated here and it's pretty cool. 00:18:58.120 --> 00:19:02.860 So what it does, it finds those dates and then you can, and then it converts them to 00:19:02.860 --> 00:19:03.400 date times. 00:19:03.400 --> 00:19:06.020 So find dates will find them and convert them to date times. 00:19:06.140 --> 00:19:09.380 And it does that by passing them off to the date util library. 00:19:09.380 --> 00:19:14.660 So, this is just kind of a really cool demo that the list, the little video is 00:19:14.660 --> 00:19:19.280 a good demo, of showing, showing how to, how to do this. 00:19:19.280 --> 00:19:21.800 I also really kind of liked this way to play. 00:19:21.800 --> 00:19:26.940 So the video shows this way to play with things of, of it just had a list of strings 00:19:26.940 --> 00:19:32.960 and then, used a comprehension to convert that, to, to call a function on a whole bunch 00:19:32.960 --> 00:19:33.380 of strings. 00:19:33.380 --> 00:19:37.920 And I thought this was just kind of a clever way to just play with a function that translates 00:19:37.920 --> 00:19:38.400 things. 00:19:38.400 --> 00:19:39.880 This is a neat thing to do. 00:19:39.880 --> 00:19:41.520 I would have probably so hard. 00:19:41.520 --> 00:19:41.820 Yeah. 00:19:41.820 --> 00:19:45.100 It's super hard, but, normally because it's so picky, right? 00:19:45.100 --> 00:19:50.240 You've got to go to the date time parsing language, almost look up. 00:19:50.320 --> 00:19:55.020 So if I put percent D, D, D, D, D, that might mean the year, but if it's capital D, it might 00:19:55.020 --> 00:19:55.880 mean something different. 00:19:55.880 --> 00:20:01.280 So you might say month, day, comma year, but like, there's an example here with month, day, 00:20:01.280 --> 00:20:04.320 year without the comma is like March 12, 2010. 00:20:04.320 --> 00:20:06.460 But if they forget the comma, it won't parse. 00:20:06.460 --> 00:20:10.300 And like all those things are really annoying about working with reading, converting strings 00:20:10.300 --> 00:20:10.640 to dates. 00:20:10.920 --> 00:20:13.080 And this looks like it just, it doesn't care. 00:20:13.080 --> 00:20:13.580 It's nice. 00:20:13.580 --> 00:20:14.180 Yeah. 00:20:14.180 --> 00:20:18.140 And then it also, it's kind of a clean, nice, clean air face to it as well. 00:20:18.140 --> 00:20:22.660 the, and a limited documentation is just a focus tool, which is nice. 00:20:23.020 --> 00:20:27.540 And it's interesting that this is just a focus tool that apparently a lot of people need 00:20:27.540 --> 00:20:31.580 because according to GitHub, there's 662 projects using this. 00:20:31.580 --> 00:20:34.340 So, it's used kind of all over the place. 00:20:34.340 --> 00:20:40.000 The behind the scenes though, it's taking the, what the dates that it found to the strings 00:20:40.000 --> 00:20:41.880 and passing those to date util. 00:20:41.880 --> 00:20:47.940 So if you want to avoid the finding part, this actually is also a good library to look at for 00:20:47.940 --> 00:20:52.860 the usage of how to use date util to easily convert, convert dates. 00:20:52.860 --> 00:20:56.200 and date util is kind of an amazing tool as well. 00:20:56.200 --> 00:21:00.800 and it gives you a, I told you this was a rabbit hole. 00:21:00.800 --> 00:21:05.520 One of the cool things about it is you, it isn't just parse dates, but you can do relative 00:21:05.520 --> 00:21:05.900 dates. 00:21:05.900 --> 00:21:10.280 You can say like today plus three weeks or something, and it'll figure that out. 00:21:10.280 --> 00:21:16.180 and then you can, or you can take two days, two dates and do, do date math with 00:21:16.180 --> 00:21:16.840 it really well. 00:21:16.840 --> 00:21:21.460 And also date util has an amazing time zone support, probably the best in Python. 00:21:21.720 --> 00:21:24.260 So, this is pretty, pretty kind of cool. 00:21:24.260 --> 00:21:27.020 also I think I was looking through the test code. 00:21:27.020 --> 00:21:33.860 The test code for date util, has, is an, is kind of a neat mix of a unit test and pie 00:21:33.860 --> 00:21:34.120 test. 00:21:34.120 --> 00:21:36.560 Both of them are good examples of how to do both. 00:21:36.560 --> 00:21:41.940 and I like some of the newer stuff is using pie test with parameterization, but it's 00:21:41.940 --> 00:21:42.140 good. 00:21:42.140 --> 00:21:42.640 Yeah. 00:21:42.640 --> 00:21:43.580 I like this a lot. 00:21:43.580 --> 00:21:44.520 Marlene, what do you think? 00:21:44.520 --> 00:21:45.680 Yeah, I like it. 00:21:45.720 --> 00:21:48.800 I think it, I'm not actually working with dates quite often. 00:21:48.800 --> 00:21:54.840 So I'm, I'm trying to think of use cases for myself other than like maybe converting time 00:21:54.840 --> 00:21:56.540 zones, which is like a nightmare. 00:21:56.540 --> 00:22:00.520 so maybe, Oh, you can say that again. 00:22:00.520 --> 00:22:01.260 Oh my gosh. 00:22:01.260 --> 00:22:07.280 Maybe you said that, but it looks like it would be really useful for people that are, yeah. 00:22:07.280 --> 00:22:07.320 Yeah. 00:22:07.320 --> 00:22:08.420 Yeah. 00:22:10.020 --> 00:22:14.160 I'm showing up some of the, some of the examples from date util of how to use it. 00:22:14.160 --> 00:22:20.320 And it's based, I imagine this is one of the reasons why date finder is so used because, 00:22:20.320 --> 00:22:23.180 um, this is non-trivial even to use date util. 00:22:23.180 --> 00:22:25.560 So yeah, that's cool. 00:22:25.560 --> 00:22:26.060 Cool. 00:22:26.060 --> 00:22:26.220 Cool. 00:22:26.220 --> 00:22:26.500 All right. 00:22:26.500 --> 00:22:31.720 Well, I got the next one and this one doesn't exactly come to us from Anthony Shaw, but I was 00:22:31.720 --> 00:22:35.600 talking to Anthony about something else and he's like, Oh, have you heard of this? 00:22:35.600 --> 00:22:37.180 Have you heard of cinder? 00:22:37.640 --> 00:22:39.940 And, cinder is pretty awesome. 00:22:39.940 --> 00:22:45.340 So Anthony's doing interesting work around Python and performance at the CPython level, 00:22:45.340 --> 00:22:49.660 especially now, I think he's given a talk on pigeon or piston, piston. 00:22:49.660 --> 00:22:50.440 I believe it is. 00:22:50.440 --> 00:22:51.980 I'm not a hundred percent sure. 00:22:51.980 --> 00:22:55.700 I might be remembering which one's wrong at PyCon, which is, you know, we're going to talk 00:22:55.700 --> 00:22:57.240 more about that in just a second as well. 00:22:57.240 --> 00:23:03.280 But cinder is a really interesting fork of CPython from Instagram. 00:23:03.280 --> 00:23:06.000 So it's under the Facebook incubator project. 00:23:06.500 --> 00:23:08.540 And I think we've mentioned it before. 00:23:08.540 --> 00:23:12.520 I definitely have talked about it before other presentations that Instagram has done really 00:23:12.520 --> 00:23:16.960 interesting things like disable the garbage collector, just turn it off a hundred percent. 00:23:16.960 --> 00:23:23.260 And they got less memory usage, not more memory usage by just allowing the cycles to leak, which 00:23:23.260 --> 00:23:23.820 is insane. 00:23:23.820 --> 00:23:27.520 But this is like, speaking of insane, this takes it to a whole nother level. 00:23:27.520 --> 00:23:31.680 So this is that they've been doing all these low level things inside of CPython. 00:23:31.800 --> 00:23:33.100 This is based on three, eight. 00:23:33.100 --> 00:23:37.640 Hopefully some of these ideas can be brought forward and shared with everyone because there's 00:23:37.640 --> 00:23:38.340 a lot going on. 00:23:38.340 --> 00:23:39.920 So let me just cruise down here. 00:23:39.920 --> 00:23:44.500 I'll just read the little intro part because it's jam packed. 00:23:44.500 --> 00:23:45.600 And then I'll go into some of the details. 00:23:45.840 --> 00:23:50.820 So this is the internal performance oriented production version of CPython 3.8. 00:23:50.820 --> 00:23:53.540 And it contains a number of performance optimizations. 00:23:53.540 --> 00:23:56.680 I feel like performance is some sort of theme of this episode. 00:23:56.680 --> 00:24:04.400 It includes bytecode inline caching, eager evaluations of coroutines, a JIT, just in time compiler, 00:24:04.920 --> 00:24:10.100 an experimental bytecode compiler that uses type annotations in some incredibly interesting 00:24:10.100 --> 00:24:14.760 ways to enter, to emit type specialized bytecode that performs better. 00:24:14.760 --> 00:24:22.220 So just to give you an example, one of the reasons that math in the pure Python layer is slower 00:24:22.220 --> 00:24:27.860 than say C++ or C# is C++ and C# work with just the value. 00:24:27.860 --> 00:24:32.000 So if you have the value seven, you might have two or four bytes that represent the value seven 00:24:32.000 --> 00:24:38.340 in Python, you have a py object pointer, which is like 28 bytes pointing out to a thing on 00:24:38.340 --> 00:24:40.440 the heap that represents the number seven. 00:24:40.440 --> 00:24:45.280 And it's a whole lot more work to interact with that and set the reference count on that 00:24:45.280 --> 00:24:47.940 and so on instead of just working with the value seven. 00:24:47.940 --> 00:24:48.440 Right. 00:24:48.440 --> 00:24:54.980 So one of the things they do is they actually have typed the, they use Python type annotations 00:24:54.980 --> 00:24:56.720 to understand, oh, this is an integer. 00:24:56.720 --> 00:25:01.620 This is a long and so on type of thing and actually convert those to the, 00:25:01.780 --> 00:25:03.640 the machine oriented numbers. 00:25:03.640 --> 00:25:03.960 Right. 00:25:03.960 --> 00:25:05.760 So just the value four instead of a pointer. 00:25:05.760 --> 00:25:07.500 And then it will use what's called boxing. 00:25:07.500 --> 00:25:12.880 If something else that's outside of this world needs it, it'll up level that to like a py long 00:25:12.880 --> 00:25:15.740 object pointer type thing and hand it off. 00:25:15.740 --> 00:25:17.920 So there's, there's all sorts of stuff like that going on. 00:25:17.920 --> 00:25:20.740 Interestingly, the first question is, is this supported? 00:25:20.740 --> 00:25:22.460 No, not supported. 00:25:26.460 --> 00:25:28.360 But there's some interesting things going on here. 00:25:28.360 --> 00:25:33.820 And all of this has to be taken within, with an understanding that it's in a very specific 00:25:33.820 --> 00:25:36.420 context and that may or may not be useful for you. 00:25:36.420 --> 00:25:40.640 You know, Brian had pointed out some articles and ideas around that. 00:25:40.700 --> 00:25:41.320 You're not Instagram. 00:25:41.320 --> 00:25:42.680 You're not Facebook. 00:25:42.680 --> 00:25:44.320 You're not Netflix. 00:25:44.320 --> 00:25:44.940 And so on. 00:25:44.940 --> 00:25:49.380 Most of the time people are building much smaller software with different constraints. 00:25:49.560 --> 00:25:54.720 So they start out by saying, look, Instagram uses a multi-process web server architecture 00:25:54.720 --> 00:26:00.620 where the parent process starts, performs initialization, and then forks 10 worker processes 00:26:00.620 --> 00:26:01.540 to handle requests. 00:26:01.540 --> 00:26:03.060 This is super common. 00:26:03.060 --> 00:26:06.080 Like for example, Talk Python training literally does exactly this. 00:26:06.080 --> 00:26:07.700 It, it uses micro-wisgi. 00:26:07.700 --> 00:26:11.420 It starts up and it creates 10 worker processes to handle like people wanting to take courses. 00:26:11.620 --> 00:26:15.980 So it's not uncommon in the web, but it's, it's not how all Python code runs. 00:26:15.980 --> 00:26:21.460 And so the first optimization they did is they created what are called immortal instances. 00:26:21.460 --> 00:26:26.420 The reason they were so focused on the garbage collector and all those sorts of things was 00:26:26.420 --> 00:26:32.280 when you fork these processes, initially there's a bunch of memory that can be shared and that 00:26:32.280 --> 00:26:37.140 helps with cache locality that helps with overall memory usage, all sorts of things. 00:26:37.540 --> 00:26:42.380 But as soon as something has changed about one of those items, it has to copy a whole page 00:26:42.380 --> 00:26:42.740 of memory. 00:26:42.740 --> 00:26:49.880 And they realized that when an object's reference count is modified in one of the processes, it 00:26:49.880 --> 00:26:55.340 has to copy a replicate and sort of fork off a bunch of the memory that used to be shared 00:26:55.340 --> 00:26:56.400 across all those processes. 00:26:56.400 --> 00:27:02.240 So they created what they call immortal instances that cannot be, that don't participate in reference 00:27:02.240 --> 00:27:03.720 counting or garbage collection. 00:27:03.720 --> 00:27:07.320 And that prohibits their reference count number to change so they can be shared. 00:27:07.480 --> 00:27:12.020 So they can mark like a whole bunch of the startup stuff as like, just don't even look at this 00:27:12.020 --> 00:27:13.880 or change it and don't do reference counting on it. 00:27:13.880 --> 00:27:17.820 So in their world, they may, it got things faster, but it doesn't always, they said it's 00:27:17.820 --> 00:27:22.200 something a little bit slower in straight line code, but in this sort of fork world, it's 00:27:22.200 --> 00:27:22.380 better. 00:27:22.380 --> 00:27:27.340 The next one is shadow byte code, which is an inline caching implementation. 00:27:28.000 --> 00:27:34.800 And it goes through applies in certain optimization cases for generic Python op codes. 00:27:34.800 --> 00:27:41.000 And it'll observe those for functions that take a lot of time and dynamically replace those 00:27:41.000 --> 00:27:44.940 with specialized op codes that it thinks are going to be better. 00:27:44.940 --> 00:27:49.660 Another thing it does that's pretty interesting is it will eagerly evaluate coroutine. 00:27:49.660 --> 00:27:54.520 So if I say this is an async method, and then in that method, I call a weight, some function 00:27:54.520 --> 00:27:58.000 call normal Python is going to create a coroutine. 00:27:58.000 --> 00:28:01.540 It's going to schedule it on the asyncio event loop, and it's going to get to it. 00:28:01.540 --> 00:28:02.760 And that's a lot of overhead. 00:28:02.760 --> 00:28:05.380 But maybe that function says inside. 00:28:05.380 --> 00:28:11.020 The first thing is if this case just return the cached answer, otherwise go to the database, 00:28:11.020 --> 00:28:13.100 await the response and so on. 00:28:13.100 --> 00:28:17.820 And what they realized is if it's going to go through that first case, it's not actually 00:28:17.820 --> 00:28:18.620 awaiting something. 00:28:18.620 --> 00:28:25.240 So they'll actually execute the awaited thing up until it actually needs to become async. 00:28:25.240 --> 00:28:29.760 So it'll like look or effectively look inside the function and say, is the path we're going 00:28:29.760 --> 00:28:31.900 on this time going to be async or not? 00:28:31.900 --> 00:28:37.300 And if the answer is no, it will run it without async, which means it skips all that context switching 00:28:37.300 --> 00:28:38.640 and all that stuff, which is pretty crazy. 00:28:38.640 --> 00:28:44.940 It also has the cinder JIT, which is a method in time JIT compiler. 00:28:44.940 --> 00:28:48.980 I think C#, Java, maybe even JavaScript V8. 00:28:48.980 --> 00:28:52.720 So it's enabled for every function that is called. 00:28:52.720 --> 00:28:54.560 Actually, it's not. 00:28:54.560 --> 00:28:54.740 Sorry. 00:28:54.740 --> 00:28:55.960 If it is, it'll make it slow. 00:28:55.960 --> 00:28:59.540 So you can basically say which functions should be optimized. 00:28:59.540 --> 00:29:03.840 But they say it supports almost everything that Python can do. 00:29:03.840 --> 00:29:09.140 And it has a 1.5 to 4 times speed up of the Python benchmarks, which is pretty interesting. 00:29:09.140 --> 00:29:15.700 They also have this thing called strict modules, which is actually a static analyzer capable of 00:29:15.700 --> 00:29:20.820 validating top level code to see if a module has side effects and can treat it differently. 00:29:20.820 --> 00:29:27.740 If it doesn't, you can have an immutable strict module type that is sort of a replacement for 00:29:27.740 --> 00:29:31.880 Python's regular module that behaves and loads differently and so on. 00:29:32.340 --> 00:29:36.820 And then the thing I talked about, the numbers more broadly is under this category of static 00:29:36.820 --> 00:29:37.200 Python. 00:29:37.200 --> 00:29:43.200 It's an experimental byte co-compiler that makes use of type annotations to emit better 00:29:43.200 --> 00:29:43.540 things. 00:29:43.540 --> 00:29:44.860 And check this out. 00:29:44.860 --> 00:29:48.180 It can deliver performance similar to mypyC or Cython. 00:29:48.180 --> 00:29:55.540 And this thing will go up to seven times faster than regular Python for the Richards benchmarks. 00:29:55.540 --> 00:29:59.100 And I don't know if the 4x improvement before is like in addition to this. 00:29:59.100 --> 00:30:01.280 So you get 28 or you just get seven. 00:30:01.360 --> 00:30:02.040 I don't really know. 00:30:02.040 --> 00:30:07.620 But there's a lot of things going on here and a lot of different ideas about how this 00:30:07.620 --> 00:30:07.920 works. 00:30:07.920 --> 00:30:12.980 So I'm just scratching the surface on the details, but I feel like I've gone on and on about it. 00:30:12.980 --> 00:30:14.340 That's really interesting. 00:30:14.340 --> 00:30:17.800 I saw, I think, is there a talk about it at Python? 00:30:17.800 --> 00:30:18.480 At PyCon? 00:30:18.480 --> 00:30:19.580 I think that's... 00:30:19.580 --> 00:30:20.560 It is coming up. 00:30:20.560 --> 00:30:20.900 Yes. 00:30:20.900 --> 00:30:22.680 They're going to give a talk on this at PyCon. 00:30:22.680 --> 00:30:22.940 Yeah. 00:30:22.940 --> 00:30:26.820 It was one of the talks I was looking forward to listening to. 00:30:26.820 --> 00:30:28.340 Yeah. 00:30:28.340 --> 00:30:34.700 Just because I think it's super interesting to be able to kind of play around with that they 00:30:34.700 --> 00:30:37.820 were able to kind of make their own version of Python. 00:30:37.820 --> 00:30:39.120 And it might... 00:30:39.120 --> 00:30:39.760 I don't know. 00:30:39.760 --> 00:30:41.080 Like, I think that there's... 00:30:41.080 --> 00:30:48.120 Like you mentioned, Anthony, and I also know Victor, I think, and someone else who are also 00:30:48.120 --> 00:30:51.880 working on, like, sub-interpreters and different things to make Python faster. 00:30:52.280 --> 00:30:58.140 So I'm really curious to see if, like, the core devs or people will also be, like, listening 00:30:58.140 --> 00:31:02.040 to this talk and maybe take some ideas from it. 00:31:02.040 --> 00:31:05.160 It would be really cool to kind of see. 00:31:05.160 --> 00:31:09.400 And I mean, it's always good to get speed-ups, even if they... 00:31:09.400 --> 00:31:09.800 I don't know. 00:31:09.800 --> 00:31:16.940 I don't know if it will help, like, general, like, normal Python users, but I think it's 00:31:16.940 --> 00:31:17.820 always good to look into. 00:31:17.820 --> 00:31:19.120 Yeah. 00:31:19.120 --> 00:31:19.520 Yeah. 00:31:19.520 --> 00:31:19.600 Yeah. 00:31:19.600 --> 00:31:20.220 I agree. 00:31:20.300 --> 00:31:26.800 I think some things here are absolutely transferable to regular general-purpose CPython, 00:31:26.800 --> 00:31:27.940 and some of them might not be. 00:31:27.940 --> 00:31:32.980 For example, the immortal instances, that might be a thing that just... 00:31:32.980 --> 00:31:36.820 They do that, and it makes sense for their large-scale farm of servers. 00:31:36.820 --> 00:31:43.740 But the JIT that takes the type information and does math many, many times faster, that... 00:31:43.740 --> 00:31:44.620 Everybody would want that. 00:31:44.620 --> 00:31:46.880 Like, we all work with numbers at some level or another. 00:31:46.880 --> 00:31:47.900 Right? 00:31:47.900 --> 00:31:50.620 Well, one of the things I love about the... 00:31:50.620 --> 00:31:54.000 I mean, this kind of applies to all of these sort of speed-up projects. 00:31:54.000 --> 00:31:57.680 One of the things I love about Python is just the generalness of it. 00:31:57.680 --> 00:31:59.020 You can throw it. 00:31:59.020 --> 00:32:00.620 Data structures can hold anything. 00:32:00.620 --> 00:32:02.580 So it's... 00:32:02.580 --> 00:32:09.280 But there are times where you really are using a huge array of floats or a huge array of integers 00:32:09.280 --> 00:32:12.940 or a huge array of, like, a fixed data size. 00:32:12.940 --> 00:32:15.860 Those are times where you... 00:32:15.860 --> 00:32:17.320 I don't need it to be generic. 00:32:17.320 --> 00:32:18.280 I just need it... 00:32:18.280 --> 00:32:19.520 I need it to be fast. 00:32:19.520 --> 00:32:21.800 So having something... 00:32:21.800 --> 00:32:26.280 That's the part where I think it would be interesting to pull into regular Python. 00:32:26.280 --> 00:32:30.860 But don't we get that with, like, some of the data science stuff anyway? 00:32:30.860 --> 00:32:32.980 Some of the number... 00:32:32.980 --> 00:32:34.320 With, like, NumPy and stuff? 00:32:34.320 --> 00:32:35.080 Yeah, you do. 00:32:35.080 --> 00:32:38.320 But you can't do generic programming with it, right? 00:32:38.320 --> 00:32:41.500 You do, like, sort of matrix math type of things. 00:32:41.500 --> 00:32:42.520 And this one... 00:32:42.520 --> 00:32:44.980 Like, the answer used to be, okay, well, this function is slow. 00:32:44.980 --> 00:32:48.320 This serialization deserialization section might be slow. 00:32:48.320 --> 00:32:50.440 So rewrite that in Cython, for example. 00:32:50.440 --> 00:32:57.340 And what's really cool about this is you can write regular Python and just put type annotations on it. 00:32:57.440 --> 00:32:59.240 And then it goes as fast as Cython. 00:32:59.240 --> 00:33:04.280 And you don't even have to do, like, a separate compiler, I believe, in this world, right? 00:33:04.280 --> 00:33:05.120 Because they have... 00:33:05.120 --> 00:33:06.360 The JIT just knows that. 00:33:06.360 --> 00:33:07.360 And then we'll... 00:33:07.360 --> 00:33:09.780 Like, as you run it, it'll just compile and run it. 00:33:09.780 --> 00:33:11.180 So, which is... 00:33:11.180 --> 00:33:16.640 I think it just sort of makes some of those ideas closer and more automatic for most people. 00:33:16.640 --> 00:33:23.920 I kind of think I foresee a future where we have sort of some types that affect runtime. 00:33:23.920 --> 00:33:35.120 There's, like, this tension that I sense in the Python core people of whether or not types should be just an afterthought or whether they should be really part of the runtime. 00:33:35.120 --> 00:33:41.660 And I think there are some cases where having them be part of the runtime might be a good thing. 00:33:42.020 --> 00:33:48.520 Yeah, and this is interesting because what they do is they define these static modules. 00:33:48.520 --> 00:33:51.020 And then in there, they can treat them differently. 00:33:51.020 --> 00:33:59.580 I feel like I always see on Twitter some people kind of, like, ranting about how they don't like that direction that Python is going in. 00:33:59.580 --> 00:34:03.200 Like, the idea of putting in, like, annotations and things like that. 00:34:03.280 --> 00:34:06.820 I've seen some people that are not super big fans of that. 00:34:06.820 --> 00:34:08.360 I'm not really sure why. 00:34:08.360 --> 00:34:12.280 I generally would like to understand, like, I think most people... 00:34:12.280 --> 00:34:14.180 Or not most people. 00:34:14.180 --> 00:34:19.500 But I think some people would prefer Python to maybe remain as it is. 00:34:19.500 --> 00:34:24.340 But I do think that there's, like, just having it be a bit lesser in a couple of cases would be helpful. 00:34:24.340 --> 00:34:25.560 So, I don't know. 00:34:25.560 --> 00:34:27.120 I don't know if it's in that direction. 00:34:28.140 --> 00:34:28.780 I'm with you. 00:34:28.780 --> 00:34:34.920 And one of the things they point out in this readme announcing the project is that you can still do gradual typing. 00:34:34.920 --> 00:34:37.680 So, you can, in some places, have no types. 00:34:37.680 --> 00:34:39.460 In some places, have some types. 00:34:39.460 --> 00:34:42.460 And the thing can convert and just deal with that automatically. 00:34:42.460 --> 00:34:48.640 And I think that's the reason that the types are really welcome in Python is because you can use them if you want, but you don't have to. 00:34:48.800 --> 00:34:55.100 As opposed to places like TypeScript, which said, well, JavaScript doesn't have types, so we're going to add this very strict type system. 00:34:55.100 --> 00:34:59.860 And if you don't fit it exactly, we're going to not compile and complain, and it's going to be really not good. 00:34:59.860 --> 00:35:05.820 This feels like it continues that forgiving nature of Python to let you opt into it. 00:35:05.820 --> 00:35:07.360 But if you do, it can go faster. 00:35:07.360 --> 00:35:09.840 That's the direction I'd like to see. 00:35:09.840 --> 00:35:16.400 I personally would like to see types be really a full-fledged feature of Python. 00:35:16.400 --> 00:35:17.520 I agree. 00:35:17.520 --> 00:35:23.860 I love that they're optional, but if they're there, let's see how much we can do and improve things with them, right? 00:35:23.860 --> 00:35:24.620 A hundred percent. 00:35:24.620 --> 00:35:25.540 Yeah. 00:35:25.540 --> 00:35:26.180 All right. 00:35:26.180 --> 00:35:27.200 Marlene, you got the last one. 00:35:27.200 --> 00:35:28.240 I got it on screen for you. 00:35:28.240 --> 00:35:28.680 Okay. 00:35:28.680 --> 00:35:29.120 Yes. 00:35:29.120 --> 00:35:34.860 The last one for today is PyCon US, which I'm very excited about. 00:35:34.860 --> 00:35:39.060 It started today, which is really great. 00:35:39.060 --> 00:35:41.440 Are both of you attending? 00:35:41.440 --> 00:35:42.380 I don't know if you're attending. 00:35:42.380 --> 00:35:45.760 Yes, absolutely. 00:35:45.760 --> 00:35:46.000 Okay. 00:35:46.000 --> 00:35:48.220 Brian, are you attending? 00:35:48.220 --> 00:35:48.740 Yay. 00:35:48.740 --> 00:35:50.260 Yeah. 00:35:50.260 --> 00:35:58.660 I think it's such a great event in terms of the fact that I know it's PyCon US, but it is, at the moment, it's the largest Python. 00:35:58.660 --> 00:36:08.440 Python gathering, or largest Python on Earth, I think, which is very cool because it means that you can meet people from all around the world. 00:36:08.440 --> 00:36:19.840 I remember, I'm really sad that it's not in person because last year, not last year, but the year before that, that's where I actually met you, Michael, for the first time. 00:36:19.840 --> 00:36:27.220 I think we were literally, I think we were at a table with you and Anthony Shaw and Łukasz Langa. 00:36:27.220 --> 00:36:29.800 And I was just randomly there. 00:36:29.800 --> 00:36:32.580 But it was such a cool discussion. 00:36:32.580 --> 00:36:40.780 And I really love the idea of being able to be in a room with people that are contributing to Python. 00:36:40.780 --> 00:36:43.740 That's my favorite part of PyCon. 00:36:43.740 --> 00:36:44.080 Yeah. 00:36:44.080 --> 00:36:45.580 It was so nice to meet you as well. 00:36:45.800 --> 00:36:48.280 That is actually my favorite part of PyCon. 00:36:48.280 --> 00:36:48.580 Yeah. 00:36:48.580 --> 00:36:54.800 It's the just, you happen to end up at a table or out for a beer or coffee with this group of people. 00:36:54.800 --> 00:36:59.240 And you're like, wow, I got these connections and this experience that just, I wouldn't. 00:36:59.240 --> 00:37:01.460 So I'm very much looking forward to coming back in person. 00:37:01.580 --> 00:37:04.080 But there's a bunch of great talks coming up. 00:37:04.080 --> 00:37:04.480 Exactly. 00:37:04.880 --> 00:37:11.000 So this year, although it's online, the online platform is very cool. 00:37:11.000 --> 00:37:13.200 And there's still lots of great talks to watch. 00:37:13.200 --> 00:37:17.460 In the show notes, I put down a list of the talks that I'm excited to watch. 00:37:17.600 --> 00:37:23.300 But I also want to just put in a word for the things that I will be doing at PyCon US this year. 00:37:23.300 --> 00:37:29.340 And the first thing I'm going to be doing is I'm going to be hosting the diversity and inclusion work group discussion, 00:37:29.340 --> 00:37:36.560 along with four other really amazing women that are part of the diversity and inclusion work group. 00:37:36.560 --> 00:37:43.220 I do want to comment here because I got like, we got some comments about it, some feedback. 00:37:43.860 --> 00:37:49.940 I posted a picture of like a group that's going to be having this discussion or hosting this panel and it's all women. 00:37:49.940 --> 00:37:52.340 And someone was just like, why is it all women? 00:37:52.340 --> 00:37:53.440 How is this diversity? 00:37:53.440 --> 00:37:57.040 So I do want to throw it out there. 00:37:57.040 --> 00:38:03.020 I just want to throw it out there that we did try, like the work group itself has a lot of, 00:38:03.020 --> 00:38:05.780 it has a good balance of men and women in it. 00:38:05.780 --> 00:38:11.240 But then when I asked people if they want to come on the panel, it was only like women that volunteered. 00:38:11.240 --> 00:38:12.280 So it's not my fault. 00:38:12.280 --> 00:38:14.300 And I am aware of that. 00:38:14.300 --> 00:38:18.440 That's just a general or that's just general feedback there. 00:38:18.440 --> 00:38:24.260 But I think the panel will be really exciting. 00:38:24.260 --> 00:38:28.760 It's going to be on Saturday on the main stage at 12 p.m. 00:38:28.760 --> 00:38:30.900 EST, I think. 00:38:30.900 --> 00:38:34.120 If you're going to be there, I really would encourage you to attend. 00:38:34.120 --> 00:38:35.660 There's going to be questions and answers. 00:38:36.420 --> 00:38:38.600 And I just think it's such an important thing. 00:38:38.600 --> 00:38:45.960 I know that sometimes diversity can seem like a really tiring thing to talk about, especially like recently. 00:38:45.960 --> 00:38:54.140 I feel like sometimes people use it as like this buzzword and it can, and people can be like, oh my gosh, and just turn off when they hear the word diversity. 00:38:54.800 --> 00:38:56.840 But I really do think it's important. 00:38:56.840 --> 00:39:10.260 And particularly now as Python is growing in popularity, I think a few years ago, it was okay for the nucleus of Python to be based in the United States or based in Europe. 00:39:10.260 --> 00:39:12.880 But it's growing so quickly. 00:39:12.880 --> 00:39:18.200 Python for I don't know how many years now has been the most popular language in the world. 00:39:18.200 --> 00:39:25.920 And I know even for me, I'm in Zimbabwe right now, and it's one of the most popular languages here where I live. 00:39:25.920 --> 00:39:42.880 And so just providing the group, like our main purpose is to figure out how we can support the PSF to try and serve Pythonistas from around the world better and to connect the community better and have better representation and different things like that. 00:39:42.880 --> 00:39:44.760 So very excited about that one. 00:39:44.760 --> 00:39:46.060 That's awesome. 00:39:46.060 --> 00:39:47.740 And thanks for your work here. 00:39:47.740 --> 00:39:51.340 I definitely agree that we're stronger together, right? 00:39:51.340 --> 00:40:00.800 And one thing I would really like to see, and I think we're getting there, is when people look at Python and programming in general, but generally the Python space, we have influence over that. 00:40:00.800 --> 00:40:06.040 When people look at that world, I would like them to say, I can see myself being part of that. 00:40:06.040 --> 00:40:08.120 I can see that I could belong there, right? 00:40:08.120 --> 00:40:11.300 And if that's not the case, then how do we make that the case? 00:40:11.300 --> 00:40:12.040 Exactly. 00:40:12.040 --> 00:40:13.020 Absolutely. 00:40:13.020 --> 00:40:15.060 I think exactly that. 00:40:15.780 --> 00:40:19.180 And I would love to see that happening in the next two years. 00:40:19.180 --> 00:40:27.220 I would love to see, you know, one of my things is I'd love to see more like women core developers and more like global core developers as well. 00:40:27.220 --> 00:40:29.800 And also people on the board and different things. 00:40:29.800 --> 00:40:31.820 And those are all goals that we are working towards. 00:40:31.960 --> 00:40:37.460 Obviously, we don't know, like, the perfect way to achieve something or the perfect way to do things. 00:40:37.460 --> 00:40:42.220 But it's something that I think is really great and exciting to work on. 00:40:42.220 --> 00:40:45.580 So please attend if you are listening to this. 00:40:46.340 --> 00:40:49.420 And let me know if you, like, came from this podcast. 00:40:49.420 --> 00:40:51.460 It would be fantastic to see you there. 00:40:51.460 --> 00:40:52.320 Maybe just comment. 00:40:52.320 --> 00:40:53.380 Yeah, fantastic. 00:40:53.380 --> 00:41:04.960 And then, oh, another thing that I am doing for, like, on this year as well is I will be, one, I will be in the, so there's, like, a lounge area. 00:41:05.160 --> 00:41:07.200 Well, there's, like, a PSF booth. 00:41:07.200 --> 00:41:15.120 And if you would like to just, if you're going to be there in the morning on Saturday or on Friday, I will be hanging out in the PSF booth. 00:41:15.640 --> 00:41:20.100 And so, yeah, if you just want to talk about Python or the PSF or anything, I will be there. 00:41:20.100 --> 00:41:24.460 And I will also be hosting the EMEA meeting. 00:41:24.460 --> 00:41:30.280 So if you're in Europe, the Middle East, or Africa, there's a members meeting on Saturday. 00:41:30.280 --> 00:41:34.000 I think it's at 10 a.m. Central African time. 00:41:34.000 --> 00:41:36.260 I'm not sure what time that is in other places. 00:41:36.260 --> 00:41:38.620 But I know it's at 10 a.m. 00:41:38.620 --> 00:41:40.100 It's on the schedule, right? 00:41:40.100 --> 00:41:40.520 It's on the schedule. 00:41:40.520 --> 00:41:40.680 Yeah. 00:41:40.680 --> 00:41:42.680 We can use the daytime thing. 00:41:42.680 --> 00:41:43.500 I don't know. 00:41:43.500 --> 00:41:45.360 Exactly. 00:41:45.880 --> 00:41:46.680 Pull up the ripple. 00:41:46.680 --> 00:41:47.640 Throw it into daytime. 00:41:47.640 --> 00:41:48.220 Exactly. 00:41:48.220 --> 00:41:49.080 Please do that. 00:41:49.080 --> 00:41:52.700 So I will be hosting that. 00:41:52.700 --> 00:41:54.600 And that's going to be in the morning. 00:41:54.600 --> 00:42:00.060 And if you would like, even if you're not a member, you can watch it on the PSF YouTube channel. 00:42:00.060 --> 00:42:01.340 It's going to be streaming there. 00:42:01.340 --> 00:42:02.540 Or you could join. 00:42:02.540 --> 00:42:04.640 There's a meetup link that I put in the show notes. 00:42:04.640 --> 00:42:06.760 So people could join that way as well. 00:42:06.760 --> 00:42:09.760 So, yeah, Python is going to be really exciting. 00:42:09.760 --> 00:42:11.420 And I'm really looking forward to it. 00:42:11.420 --> 00:42:14.400 So just encouraging people to come along for sure. 00:42:14.680 --> 00:42:15.480 Yeah, it should be fun. 00:42:15.480 --> 00:42:21.440 And even though it is super sad that it's not in person, it's not in Pittsburgh this year. 00:42:21.440 --> 00:42:25.100 I think in some ways it's more accessible to people around the world, right? 00:42:25.100 --> 00:42:26.340 They don't have to travel there. 00:42:26.340 --> 00:42:29.140 They can just log in and attend it. 00:42:29.140 --> 00:42:32.680 And that's so much less expensive than I flew to the U.S. 00:42:32.680 --> 00:42:34.880 And I paid $1,000 for a hotel. 00:42:34.880 --> 00:42:38.980 So there's a little silver lining, you know, out there in the live stream. 00:42:38.980 --> 00:42:42.520 Sam Morley really says, I really wish I could go to PyCon in person. 00:42:42.520 --> 00:42:45.940 Adam Parkin there says, me too, maybe in 2020. 00:42:45.940 --> 00:42:46.920 I think so. 00:42:46.920 --> 00:42:53.220 Finally, Sam also thinks it's great that we're having this diversity conversation and paying attention to it. 00:42:53.340 --> 00:43:00.000 One of the things I've noticed in 2020 is all the regional, actually last year also, though. 00:43:00.100 --> 00:43:07.400 But the 2020 and 2021, we've got all these PyCons going on all over the world. 00:43:07.400 --> 00:43:12.760 I used to think of like PyCon U.S. as the PyCon and everything else is regional. 00:43:13.200 --> 00:43:17.680 Now I think of PyCon U.S. as a regional conference also. 00:43:17.680 --> 00:43:21.320 It's the regional one that's close to the people that are in the U.S. 00:43:21.320 --> 00:43:23.660 It isn't necessarily better. 00:43:23.660 --> 00:43:26.020 It's, I love it. 00:43:26.020 --> 00:43:26.520 It's great. 00:43:26.520 --> 00:43:29.320 Anybody from that's hosting it, yes, it's better. 00:43:29.320 --> 00:43:34.080 But no, I like all of them. 00:43:34.080 --> 00:43:39.680 And I was excited to get to participate and watch videos from all over the world this year. 00:43:39.680 --> 00:43:40.840 That was pretty neat. 00:43:41.360 --> 00:43:45.980 But yeah, I'm on board with, I want to get back to regional stuff. 00:43:45.980 --> 00:43:47.940 I'd like to see people in person. 00:43:47.940 --> 00:43:48.980 I can't wait. 00:43:48.980 --> 00:43:56.320 Yeah, I will say for sure, like, even if people are feeling adventurous, there is a regional conference. 00:43:56.320 --> 00:44:01.740 I didn't mention it before that I am also part of the organizing team for, which is PyCon Africa. 00:44:01.740 --> 00:44:10.820 So if you would like to travel to another PyCon in a different part of the world, when we are able to travel and the world gets back to, 00:44:11.020 --> 00:44:19.060 some form of free travel, definitely recommend also hopping over to PyCon Africa. 00:44:19.060 --> 00:44:23.400 I think, like you said, I think PyCon US is fantastic. 00:44:23.400 --> 00:44:29.960 And one of the neat things about that is that it's a conference that has been there for so long. 00:44:29.960 --> 00:44:31.760 So a lot of people are going to be there. 00:44:32.640 --> 00:44:40.600 But there are 100% are a lot of great conferences like PyCon Africa, which you should attend if you can. 00:44:40.800 --> 00:44:42.540 I think they're really just as exciting. 00:44:42.540 --> 00:44:45.800 And there's so many cool things that you get to experience. 00:44:45.800 --> 00:44:51.840 Like, I think for me, it's like whenever I go to the US, like last year, I'd never been to Ohio before. 00:44:51.840 --> 00:45:02.560 And like, I had like, I would never like in my, I would never have a reason that I would think to myself, let me go to America to go to Ohio. 00:45:03.680 --> 00:45:07.460 But I feel like it was such a good experience for me. 00:45:07.460 --> 00:45:08.880 And I really liked it. 00:45:08.880 --> 00:45:10.680 And I was really surprised by that. 00:45:10.680 --> 00:45:12.580 And so I think it's the same way. 00:45:12.580 --> 00:45:15.820 Like, PyCon is a great way as well to like experience new places. 00:45:16.140 --> 00:45:18.320 So yeah, definitely sticking in that. 00:45:18.320 --> 00:45:20.580 Well, that wraps up our six. 00:45:20.580 --> 00:45:24.440 Anybody got any extra information to share? 00:45:24.440 --> 00:45:31.540 Nothing else for me, other than the fact that if you do want to reach out to me, you can reach out to me on Twitter. 00:45:31.540 --> 00:45:34.480 I'm Marlene underscore ZW there. 00:45:34.480 --> 00:45:38.300 I'm also Marlene underscore ZW on GitHub, I think. 00:45:38.300 --> 00:45:40.500 And on my website. 00:45:40.500 --> 00:45:42.880 My website is MarleneMangami.com. 00:45:43.120 --> 00:45:46.700 So if you would like to reach out to me here, feel free to. 00:45:46.700 --> 00:45:49.120 I'm always happy to like chat about PyCon. 00:45:49.120 --> 00:45:50.320 Nice. 00:45:50.320 --> 00:45:50.840 Cool. 00:45:50.840 --> 00:45:51.780 So I got a couple. 00:45:51.780 --> 00:45:53.200 One made me really excited. 00:45:53.200 --> 00:45:55.040 This tweet from GitHub. 00:45:55.040 --> 00:45:57.040 Is your fork behind? 00:45:57.040 --> 00:46:01.640 You can now sync your parent repo with just a single click. 00:46:01.640 --> 00:46:02.620 So check this out. 00:46:02.620 --> 00:46:09.960 If you go to your fork now, next to contribute for your PRs and stuff, there is now a fetch upstream button. 00:46:10.100 --> 00:46:16.420 And all you have to do is click it and then automatically your fork will become in sync with whatever you forked it from. 00:46:16.420 --> 00:46:19.020 You just have to go and go check it out. 00:46:19.020 --> 00:46:24.160 Add an upstream origin and then pull from that and then merge that wherever you want it to go to. 00:46:24.160 --> 00:46:27.040 Over here, you just click this button and boom. 00:46:27.040 --> 00:46:27.840 It's good to go. 00:46:28.180 --> 00:46:31.320 So I think this will just lower the bar for people forking something. 00:46:31.320 --> 00:46:35.300 They want to get the current one and then make a change to see if they could contribute back. 00:46:35.300 --> 00:46:37.880 Here's one fewer steps in that process. 00:46:37.880 --> 00:46:41.520 Do you have any idea if it stays in sync or if you have to? 00:46:41.520 --> 00:46:43.920 No, it's a one-time type of thing, I believe. 00:46:43.920 --> 00:46:49.380 It says there's this many changes we'll pull over and it basically just automatically does the process at that time. 00:46:49.380 --> 00:46:49.880 Nice. 00:46:49.880 --> 00:46:51.240 But still pretty nice. 00:46:51.240 --> 00:46:53.980 You know, Flask 2.0 is out. 00:46:53.980 --> 00:47:01.640 And that one was sent in to us from Adam Parkin that, hey, heads up, this is now actually live. 00:47:01.640 --> 00:47:03.100 So very, very cool. 00:47:03.540 --> 00:47:06.180 Actually, everything from Palettes has been updated. 00:47:06.180 --> 00:47:07.440 So, yeah. 00:47:07.440 --> 00:47:21.900 I happen to have spoken, done a podcast recording with David Lord, who runs Palettes, and Phil Jones, who does Core and contributes back to Palettes as well, about all the stuff coming in Flask 2.0, all the exciting stuff and their future plans as well. 00:47:21.900 --> 00:47:23.040 So, yeah. 00:47:23.040 --> 00:47:29.080 You can watch the live stream of that or wait a day or two until the episode is out and just listen at Talk Python as well. 00:47:29.080 --> 00:47:30.240 But, yeah, very, very cool. 00:47:30.240 --> 00:47:30.580 Yeah. 00:47:30.960 --> 00:47:33.760 And then Adam also at the live stream again says, this is super sweet. 00:47:33.760 --> 00:47:35.880 Always find it a headache to sync with upstream. 00:47:35.880 --> 00:47:37.080 Yeah, about the GitHub thing. 00:47:37.080 --> 00:47:37.860 That's cool. 00:47:37.860 --> 00:47:38.400 Cool. 00:47:38.400 --> 00:47:39.460 Close it out with a joke. 00:47:39.460 --> 00:47:41.960 Well, I got a couple of things I wanted to mention. 00:47:41.960 --> 00:47:42.880 Go for it. 00:47:42.880 --> 00:47:43.120 Sorry. 00:47:43.120 --> 00:47:51.080 I had Brett Cannon on last week on testing code and huge feedback from everybody that it was a great episode. 00:47:51.080 --> 00:47:52.120 We talked about packaging. 00:47:52.120 --> 00:47:56.260 I'll have Ryan Howard on this week talking about Playwright. 00:47:56.260 --> 00:47:57.540 So that'll be fun. 00:47:57.960 --> 00:48:03.300 And I wanted to mention a thank you to the 71 patrons that we have on Patreon. 00:48:03.300 --> 00:48:05.300 So thank you for supporting the show. 00:48:05.300 --> 00:48:05.900 Thanks. 00:48:05.900 --> 00:48:06.480 Yeah. 00:48:06.480 --> 00:48:07.000 Thank you, everyone. 00:48:07.000 --> 00:48:08.060 How about a joke? 00:48:08.060 --> 00:48:09.040 Yes. 00:48:09.040 --> 00:48:11.820 Sorry for almost skipping over your extras. 00:48:11.820 --> 00:48:13.300 Here. 00:48:13.300 --> 00:48:14.140 No worries. 00:48:14.140 --> 00:48:15.040 You ready? 00:48:15.040 --> 00:48:15.740 Yeah. 00:48:15.800 --> 00:48:20.060 So this one, I talked about that crazy giant ship thing. 00:48:20.060 --> 00:48:22.060 And we've got Marlene doing rapid. 00:48:22.060 --> 00:48:24.840 So I thought maybe some kind of machine learning joke. 00:48:24.840 --> 00:48:29.340 Here's a bunch of robots in school. 00:48:29.340 --> 00:48:32.260 And they go like little Android looking things. 00:48:32.260 --> 00:48:34.460 Small ones because they're students. 00:48:34.460 --> 00:48:34.900 They're kids. 00:48:34.900 --> 00:48:36.640 And they're in machine learning class. 00:48:37.040 --> 00:48:38.980 And there's a big box of dirty data. 00:48:38.980 --> 00:48:42.120 Like a bunch of bits that are like kind of gray. 00:48:42.120 --> 00:48:42.980 And I don't know. 00:48:42.980 --> 00:48:43.840 They just have dirt on them. 00:48:43.840 --> 00:48:47.480 And the teacher says, Robbie, stop misbehaving. 00:48:47.480 --> 00:48:49.440 Or I will send you back to data cleaning. 00:48:49.440 --> 00:48:53.800 Yeah. 00:48:53.800 --> 00:48:56.180 That's where they're spending half the day anyway. 00:48:56.180 --> 00:48:56.600 Yeah. 00:48:56.600 --> 00:48:58.240 They actually spend most of their time there. 00:48:58.240 --> 00:48:58.660 That's right. 00:48:59.220 --> 00:49:00.740 I don't know who is drawing them. 00:49:00.740 --> 00:49:03.060 Like one of the robots is looking the wrong way. 00:49:03.060 --> 00:49:04.780 I was like, why is it drawn like that? 00:49:04.780 --> 00:49:05.500 I don't understand. 00:49:05.500 --> 00:49:12.800 Hey, a more concrete, really quick closeout question I see in the live stream here is 00:49:12.800 --> 00:49:13.440 Akmos. 00:49:13.440 --> 00:49:18.420 Is there a difference between QDF and pandas in terms of utilization? 00:49:18.420 --> 00:49:22.960 Like in terms of how you actually use them? 00:49:22.960 --> 00:49:26.740 Well, I don't think so. 00:49:26.860 --> 00:49:32.700 For the most part, if when you're using QDF, the way it's built is to mirror pandas. 00:49:32.700 --> 00:49:35.160 So the APIs are really similar. 00:49:35.160 --> 00:49:40.840 So ideally, the methods that you would use when you're using pandas are exactly the same 00:49:40.840 --> 00:49:43.420 methods that you would use when you're using QDF. 00:49:43.420 --> 00:49:47.620 The only difference is like when you're creating a pandas data frame, for example, 00:49:47.620 --> 00:49:50.860 you would use PD dot data frame, for example. 00:49:50.860 --> 00:49:54.080 But then with QDF, you would say QDF dot data frame. 00:49:54.400 --> 00:49:59.580 If you make it like into a variable or something like that, then the methods that you're going 00:49:59.580 --> 00:50:01.400 to pull are going to be totally identical. 00:50:01.400 --> 00:50:03.200 It's really easy to try. 00:50:03.200 --> 00:50:04.240 Yeah, that's it. 00:50:04.240 --> 00:50:05.000 Yeah, that's awesome. 00:50:05.000 --> 00:50:05.640 Yeah. 00:50:05.640 --> 00:50:07.500 And Dask has similar stuff as well, right? 00:50:07.500 --> 00:50:11.220 You create a Dask data frame instead of a pandas data frame, but the API looks quite similar. 00:50:11.220 --> 00:50:16.060 They're not always 100% compatible, but most of the mainstream things, right? 00:50:16.240 --> 00:50:17.120 Definitely. 00:50:17.120 --> 00:50:23.780 So yeah, it's built definitely to make it as easy as possible to switch between the two. 00:50:23.780 --> 00:50:24.560 So it's very similar. 00:50:24.560 --> 00:50:24.800 Yeah. 00:50:24.800 --> 00:50:26.640 Thanks a lot, everybody, for showing up. 00:50:26.640 --> 00:50:27.080 Yeah. 00:50:27.080 --> 00:50:27.720 Thanks. 00:50:27.720 --> 00:50:28.600 Thanks, Maya. 00:50:28.600 --> 00:50:29.400 Thank you, Marlene. 00:50:29.400 --> 00:50:30.320 It's really great to have you here. 00:50:30.320 --> 00:50:31.240 No problem. 00:50:31.240 --> 00:50:32.500 Thanks for having me. 00:50:33.020 --> 00:50:34.640 Thank you for listening to Python Bytes. 00:50:34.640 --> 00:50:37.200 Follow the show on Twitter via at Python Bytes. 00:50:37.200 --> 00:50:40.080 That's Python Bytes as in B-Y-T-E-S. 00:50:40.080 --> 00:50:43.500 And get the full show notes at pythonbytes.fm. 00:50:43.500 --> 00:50:47.840 If you have a news item you want featured, just visit pythonbytes.fm and send it our way. 00:50:47.840 --> 00:50:50.540 We're always on the lookout for sharing something cool. 00:50:50.540 --> 00:50:53.800 On behalf of myself and Brian Okken, this is Michael Kennedy. 00:50:53.800 --> 00:50:57.360 Thank you for listening and sharing this podcast with your friends and colleagues.