WEBVTT 00:00:00.001 --> 00:00:03.180 I am pulling off a very, very cool trick. 00:00:03.180 --> 00:00:04.860 I just want to point out before we get started. 00:00:04.860 --> 00:00:05.440 Okay. 00:00:05.440 --> 00:00:07.100 On the Talk Python channel, 00:00:07.100 --> 00:00:10.500 I'm doing a podcast with Anthony Shaw and Shane from Microsoft 00:00:10.500 --> 00:00:14.900 about Azure and Python and some CLI stuff they built and FastAPI. 00:00:14.900 --> 00:00:17.720 And at the exact same time, I'm doing this one here. 00:00:17.720 --> 00:00:18.700 They're both streaming live. 00:00:18.700 --> 00:00:21.900 I don't know how that's happening. 00:00:21.900 --> 00:00:25.120 The other one was recorded two months ago, 00:00:25.120 --> 00:00:27.840 and we couldn't release it because some of the things weren't finished yet. 00:00:27.840 --> 00:00:29.380 So I hit go on that. 00:00:29.680 --> 00:00:32.580 The real one, if you're bouncing around, the real one is here. 00:00:32.580 --> 00:00:33.200 Okay. 00:00:33.200 --> 00:00:34.440 So a joint is here. 00:00:34.440 --> 00:00:36.680 Anyway, with that, you ready to start a podcast? 00:00:36.680 --> 00:00:37.760 Yeah, definitely. 00:00:37.760 --> 00:00:39.560 Hello, and welcome to Python Bytes, 00:00:39.560 --> 00:00:42.780 where we deliver Python news and headlines directly to your earbuds. 00:00:42.780 --> 00:00:47.260 This is episode 294, recorded July 12th, 2022. 00:00:47.260 --> 00:00:48.260 I'm Michael Kennedy. 00:00:48.260 --> 00:00:49.720 And I am Brian Okken. 00:00:49.720 --> 00:00:51.920 It's just us this weekend, or this today. 00:00:51.920 --> 00:00:52.400 It's just us. 00:00:52.400 --> 00:00:52.860 Yeah. 00:00:52.860 --> 00:00:55.640 Yeah, it's, I don't know. 00:00:55.640 --> 00:00:58.780 Dean out of the audience asks, is this a daily podcast show now? 00:00:59.360 --> 00:01:01.560 I'm a little bit torn about it. 00:01:01.560 --> 00:01:03.320 I feel like we almost could do a daily show. 00:01:03.320 --> 00:01:06.200 But then I think what it might take to do a daily show, 00:01:06.200 --> 00:01:08.420 knowing how much work a weekly show is. 00:01:08.420 --> 00:01:10.540 No, it's not a daily podcast. 00:01:10.540 --> 00:01:10.800 No. 00:01:10.800 --> 00:01:12.460 It might be fun to do sometime. 00:01:12.460 --> 00:01:14.400 Just do like a full week or something. 00:01:14.400 --> 00:01:15.120 Right. 00:01:15.200 --> 00:01:15.560 Exactly. 00:01:15.560 --> 00:01:18.400 Just a super, there's so much news. 00:01:18.400 --> 00:01:20.060 We're doing it every day for the week. 00:01:20.060 --> 00:01:20.640 Cool. 00:01:20.640 --> 00:01:23.300 But just like the same topics, like six days in a row. 00:01:23.300 --> 00:01:23.900 Just do them over. 00:01:23.900 --> 00:01:25.120 Yeah. 00:01:25.120 --> 00:01:27.520 Exactly. 00:01:27.520 --> 00:01:28.060 Exactly. 00:01:28.060 --> 00:01:29.040 All right. 00:01:29.040 --> 00:01:30.500 Am I up first this week? 00:01:30.500 --> 00:01:31.340 You are. 00:01:31.340 --> 00:01:31.920 Yes. 00:01:31.920 --> 00:01:32.900 Right on. 00:01:32.900 --> 00:01:36.180 Well, let me tell you about something special. 00:01:36.180 --> 00:01:37.240 Specialist. 00:01:37.240 --> 00:01:37.760 Okay. 00:01:38.100 --> 00:01:44.080 Just last week, I believe it was, I interviewed Alex Waygood, who did the write-up for the Python 00:01:44.080 --> 00:01:44.760 Language Summit. 00:01:44.760 --> 00:01:49.660 And as part of the topics we were discovering, you know, the Python Language Summit and Python 00:01:49.660 --> 00:01:53.920 this year is focusing a lot on performance and what's called the Shannon Plan. 00:01:54.300 --> 00:01:59.140 So this is Mark Shannon's plan to make Python five times faster over five releases. 00:01:59.140 --> 00:02:01.700 It's got a ton of support at Microsoft. 00:02:01.700 --> 00:02:06.740 Peter Van Rossum is there working on it, but they've hired like five or six other people who 00:02:06.740 --> 00:02:08.920 are full-time working on making Python faster now. 00:02:08.920 --> 00:02:09.780 So awesome. 00:02:09.780 --> 00:02:10.140 Awesome. 00:02:10.140 --> 00:02:11.160 Thank you for that. 00:02:11.160 --> 00:02:18.020 However, one of the things that made Python 3.11 fast is some of the early work they did. 00:02:18.020 --> 00:02:24.740 And it comes down to PEP 659, a specializing adaptive interpreter. 00:02:24.740 --> 00:02:30.400 So let me tell you about this feature, this performance improvement first, and then we'll 00:02:30.400 --> 00:02:34.780 see what specialist is about because it's about understanding and visualizing this behavior. 00:02:34.780 --> 00:02:35.240 Okay. 00:02:35.240 --> 00:02:42.940 So one of the things that is a problem with Python, because it's dynamic and its types can change 00:02:42.940 --> 00:02:44.660 and what can be passed could vary. 00:02:44.660 --> 00:02:48.440 I mean, you could have type hints, but you can violate the type hints all day long and 00:02:48.440 --> 00:02:48.900 it's fine. 00:02:48.900 --> 00:02:54.500 So what the interpreter has to do is say, well, we're going to do all of our operations super 00:02:54.500 --> 00:02:54.920 general. 00:02:54.920 --> 00:02:59.820 So if I have a function and it's called add and it takes X and Y and it returns X plus Y 00:02:59.820 --> 00:03:02.760 seems easy, but is that string addition? 00:03:02.760 --> 00:03:05.280 Is that numerical addition? 00:03:05.280 --> 00:03:12.200 Is that some custom operator overloading with a dunder add or whatever it is in some type? 00:03:12.200 --> 00:03:14.220 If it fails in one way, you kind of got to reverse it. 00:03:14.220 --> 00:03:15.720 Like there's all this unknown, right? 00:03:15.720 --> 00:03:16.180 Yeah. 00:03:16.180 --> 00:03:22.420 What if you knew, what if you knew those were integers and not classes or not strings? 00:03:22.420 --> 00:03:24.040 You could run different code. 00:03:24.040 --> 00:03:26.400 You wouldn't have to first figure out what they are. 00:03:26.400 --> 00:03:27.060 Are they compatible? 00:03:27.060 --> 00:03:32.460 Do you do the add in low level CPython internals? 00:03:32.460 --> 00:03:34.420 Or do you go to like some Python class and do it? 00:03:34.420 --> 00:03:34.760 Right. 00:03:34.760 --> 00:03:36.260 You could be much more focused. 00:03:36.260 --> 00:03:36.800 Yeah. 00:03:37.220 --> 00:03:41.200 Additionally, if it was adding for a list, you could say, well, if I know their list, 00:03:41.200 --> 00:03:45.940 what we just do is go list dot extend and we give it the other list, right? 00:03:45.940 --> 00:03:48.040 We don't hunt around and figure out all this other stuff. 00:03:48.040 --> 00:03:54.020 So that's the general idea of the specializing interpreter is it goes through and it says, look, 00:03:54.020 --> 00:03:57.200 we don't know for sure what could be passed here. 00:03:57.380 --> 00:04:03.220 But if it looks like over and over we're running the same code and it's always the same types, 00:04:03.220 --> 00:04:07.900 is there a way we could specialize those types, right? 00:04:07.900 --> 00:04:14.140 Is there a way that we could put specific code for adding numbers or specific code for combining lists? 00:04:14.140 --> 00:04:18.900 And this is called adaptive and speculative specialization. 00:04:18.900 --> 00:04:19.760 Okay. 00:04:19.760 --> 00:04:20.400 Okay. 00:04:21.140 --> 00:04:25.220 And my favorite part of it, when it's performed, it's called the quickening. 00:04:25.220 --> 00:04:30.940 Quickening is the process of replacing slow instructions with faster variants. 00:04:30.940 --> 00:04:35.240 So kind of like I said, it has some advantages over immutable byte code. 00:04:35.240 --> 00:04:36.620 It can be changed at runtime. 00:04:36.620 --> 00:04:39.100 Like you see, we're always adding integers. 00:04:39.100 --> 00:04:43.460 It can use super instructions that span lines or take multiple operands. 00:04:43.460 --> 00:04:48.420 And it does not need to handle tracing as it can fall back to the original byte code for that. 00:04:48.420 --> 00:04:48.800 Okay. 00:04:48.800 --> 00:04:49.380 Okay. 00:04:49.380 --> 00:04:51.520 So there's a whole bunch of stuff going on here. 00:04:51.520 --> 00:04:56.060 Like the example they give is you might want to specialize load adder. 00:04:56.060 --> 00:05:00.360 So load adder is a way to say, give me the value that this thing contains. 00:05:00.360 --> 00:05:01.660 But what is the thing? 00:05:01.660 --> 00:05:05.840 One of the things you might do is you might realize it's an instance class. 00:05:05.840 --> 00:05:09.080 And then you would call load adder instance value. 00:05:09.080 --> 00:05:09.600 Okay. 00:05:09.600 --> 00:05:14.460 You might realize it's a module and you might call load adder module or slot or so on, right? 00:05:14.460 --> 00:05:17.620 But if you knew, you don't have to go through first the abstract step 00:05:17.620 --> 00:05:20.120 and then figure out which of these it is. 00:05:20.120 --> 00:05:21.460 You just do the thing that it is. 00:05:21.460 --> 00:05:22.080 Okay. 00:05:22.080 --> 00:05:24.340 So that's the idea of this PEP. 00:05:24.340 --> 00:05:27.720 This is one of the things that's making Python 3.11 faster. 00:05:27.720 --> 00:05:28.380 Awesome. 00:05:28.380 --> 00:05:30.180 So to the main topic. 00:05:30.180 --> 00:05:30.900 Okay. 00:05:30.900 --> 00:05:35.540 And I'll just, just as a note, I'm saying, okay, as if I understand what you just said, 00:05:35.540 --> 00:05:37.080 but most of it just went. 00:05:37.080 --> 00:05:38.240 It's all right. 00:05:38.240 --> 00:05:39.740 I think, well, let's, let's look at pictures. 00:05:39.740 --> 00:05:40.280 Okay. 00:05:40.280 --> 00:05:40.920 All right. 00:05:40.920 --> 00:05:46.300 So this thing by Brant Boucher is called Specialist. 00:05:46.300 --> 00:05:50.140 And it's about visualizing this specializing adaptive interpreter. 00:05:50.140 --> 00:05:50.880 Okay. 00:05:50.880 --> 00:05:51.240 Good. 00:05:51.240 --> 00:05:52.020 Okay. 00:05:52.020 --> 00:05:57.220 So it says Specialist uses fine-grained location information to create visual representations 00:05:57.220 --> 00:06:04.220 of exactly where and how CPython 3.11's new specializing adaptive interpreter optimizes 00:06:04.220 --> 00:06:04.740 your code. 00:06:04.740 --> 00:06:06.820 And it's not just interesting. 00:06:06.820 --> 00:06:08.720 It has actionable information. 00:06:08.720 --> 00:06:14.740 So for example, see here, and if you've got to pull up this, the website, if you're just 00:06:14.740 --> 00:06:17.880 listening, if you see in that website, you'll see some color. 00:06:18.140 --> 00:06:23.560 You'll see green, less green, yellow, orange, and all the way to red. 00:06:23.560 --> 00:06:24.560 So there's two aspects. 00:06:24.560 --> 00:06:27.420 There's sort of a darkness as well as a color. 00:06:27.420 --> 00:06:33.300 So the most, like where Python could take advantage of this feature, you see green 00:06:33.300 --> 00:06:34.960 where it can't, you see red. 00:06:34.960 --> 00:06:37.900 And imagine a spectrum. 00:06:37.900 --> 00:06:40.880 It goes like green, yellow, orange, red. 00:06:40.880 --> 00:06:42.220 So it's, it's not on or off. 00:06:42.220 --> 00:06:43.640 It's how much could it specialize? 00:06:43.640 --> 00:06:44.260 Okay. 00:06:44.260 --> 00:06:44.620 Okay. 00:06:44.620 --> 00:06:52.080 So what you see here, for example, is it's able to take some numbers and an integer and 00:06:52.080 --> 00:06:57.580 a string, and then use the fact that it knows what those are to make certain things like a 00:06:57.580 --> 00:07:02.240 pending and output and doing some character operations on it. 00:07:02.240 --> 00:07:02.740 Yeah. 00:07:02.740 --> 00:07:03.040 Right. 00:07:03.040 --> 00:07:07.080 It was able to replace that with a different runtime behavior because of this quickening. 00:07:07.080 --> 00:07:07.640 All right. 00:07:07.640 --> 00:07:08.880 So let's skip down here. 00:07:08.880 --> 00:07:10.540 I gave you a bit of the background. 00:07:10.540 --> 00:07:11.860 So it says, let's look at this example. 00:07:11.860 --> 00:07:15.680 We have F to C, which converts Fahrenheit to Celsius. 00:07:15.680 --> 00:07:20.480 And what it does is, okay, we're going to take an F and it has type hints that say float, 00:07:20.480 --> 00:07:20.780 float. 00:07:20.780 --> 00:07:21.100 Okay. 00:07:21.100 --> 00:07:22.280 So, but those don't matter. 00:07:22.280 --> 00:07:26.700 So it says, we're going to take an F and subtract 32 from it. 00:07:26.700 --> 00:07:28.420 And then we're going to do simple math. 00:07:28.420 --> 00:07:33.940 We're going to take that result, that range, that, that, size of temperature there based 00:07:33.940 --> 00:07:36.560 on zero and then multiply it by five and divide it by nine. 00:07:36.560 --> 00:07:41.520 And we all learned this in chemistry class or somewhere, or we talked about diverting different, 00:07:41.520 --> 00:07:42.840 um, measurement. 00:07:42.840 --> 00:07:43.820 Yeah, of course. 00:07:43.820 --> 00:07:44.200 Yeah. 00:07:44.200 --> 00:07:44.660 Right. 00:07:44.660 --> 00:07:50.700 So these are straightforward, but there's actually problems in here that make it slower and prohibit 00:07:50.700 --> 00:07:54.400 Python from quickening it as much as it can be quickened. 00:07:54.400 --> 00:07:55.120 Okay. 00:07:55.440 --> 00:07:59.740 So if we take this code, it just runs, F to C and C to F, and it gives it some test 00:07:59.740 --> 00:08:02.060 values and says, just do it and tell us what happened. 00:08:02.060 --> 00:08:03.760 We can run specialists on it. 00:08:03.760 --> 00:08:08.540 And it says, okay, this X here is, the green areas indicate regions of code that were 00:08:08.540 --> 00:08:13.000 successfully specialized where red areas are unsuccessful. 00:08:13.000 --> 00:08:14.520 Like it tried and it failed. 00:08:14.520 --> 00:08:20.220 So it says one of the problems is start out the X equals F minus 32. 00:08:20.220 --> 00:08:26.480 It says, well, we can quicken operations on numerical types that are the same, but for now 00:08:26.480 --> 00:08:29.560 there's not a float int and float variant of this. 00:08:29.560 --> 00:08:30.660 It's got to be float float. 00:08:30.660 --> 00:08:31.340 Oh, right. 00:08:31.340 --> 00:08:32.400 So it says, right. 00:08:32.400 --> 00:08:36.800 You, you could have gotten a faster operation there, but because the types didn't match, 00:08:36.800 --> 00:08:37.480 you won't. 00:08:37.480 --> 00:08:39.840 But then what it did get out is an X and that was great. 00:08:39.840 --> 00:08:43.020 an X, which is a float and it's going to do some stuff and it could sort of make it 00:08:43.020 --> 00:08:47.660 better, but it said, look, here's some multiplication again by an integer and a float. 00:08:47.660 --> 00:08:48.660 So that's not quickened. 00:08:48.660 --> 00:08:51.920 And this division division is apparently never quickened. 00:08:51.920 --> 00:08:52.980 So what can we do? 00:08:52.980 --> 00:08:57.180 Well, with that information to say, well, what's the problem with subtracting 32? 00:08:57.180 --> 00:08:58.360 Well, it wasn't a float. 00:08:58.360 --> 00:08:59.940 What if I said 32.0? 00:08:59.940 --> 00:09:00.920 Oh, yes. 00:09:00.920 --> 00:09:01.280 All right. 00:09:01.280 --> 00:09:02.900 That gets replaced by faster code. 00:09:02.900 --> 00:09:03.760 Oh, nice. 00:09:03.760 --> 00:09:03.960 Right. 00:09:03.960 --> 00:09:04.200 Yeah. 00:09:04.200 --> 00:09:05.260 So that's pretty nice. 00:09:05.260 --> 00:09:09.360 And if you want to return, it was adding like X plus 32 for the other direction. 00:09:09.740 --> 00:09:10.740 And now it's 32.0. 00:09:10.740 --> 00:09:11.400 That's faster. 00:09:11.400 --> 00:09:11.940 Okay. 00:09:11.940 --> 00:09:12.700 Well, what else? 00:09:12.700 --> 00:09:19.080 What if we, now you can see when we did that part of the, conversion X times five divided 00:09:19.080 --> 00:09:24.300 by nine, if we put a 5.0, that gets faster still, but the divide is never quickened. 00:09:24.300 --> 00:09:24.680 Okay. 00:09:24.680 --> 00:09:26.840 Well, what if we put the divide in parentheses? 00:09:26.840 --> 00:09:32.340 It doesn't really matter if it's X times five divided by nine or X times five divided by 00:09:32.340 --> 00:09:32.820 nine, right? 00:09:32.820 --> 00:09:36.920 It's, these are mathematically equivalent, but they're not equivalent to Python because 00:09:36.920 --> 00:09:42.820 that, that operation results in, it leverages constant folding, right? 00:09:42.820 --> 00:09:46.020 Five divided by nine is pre-computed in Python to be a float. 00:09:46.020 --> 00:09:46.640 Okay. 00:09:46.640 --> 00:09:47.120 Right. 00:09:47.120 --> 00:09:48.120 At parse time, right? 00:09:48.120 --> 00:09:49.300 That's just how it works with constants. 00:09:49.300 --> 00:09:51.880 If it says it can do math with constants ahead of time, it does it. 00:09:51.880 --> 00:09:53.040 So that becomes a float. 00:09:53.040 --> 00:09:54.780 And then float times float is now quickened. 00:09:54.780 --> 00:09:55.260 Right. 00:09:55.480 --> 00:09:59.360 Isn't this cool the way you can apply this and actually make your code faster, not just 00:09:59.360 --> 00:10:00.200 go, Oh, it's interesting. 00:10:00.200 --> 00:10:02.680 It must be quick in it there, but it's actionable. 00:10:02.680 --> 00:10:04.140 It is really pretty cool. 00:10:04.140 --> 00:10:09.420 And I'd really like to see this incorporated into an editor or something to say you, your 00:10:09.420 --> 00:10:13.300 code will be faster if you just add a point zero here or something like that. 00:10:13.300 --> 00:10:15.520 And it's going to become a float anyway. 00:10:15.520 --> 00:10:16.260 It doesn't matter. 00:10:16.260 --> 00:10:21.540 It just, why would you write 32.0 when you just meant 32 seems more precise to say 32. 00:10:22.860 --> 00:10:26.040 Cause I'm used to doing that to thinking if it's okay. 00:10:26.040 --> 00:10:30.400 Well, me personally, I, if I know it's going to be a float math, I usually do point zero, 00:10:30.400 --> 00:10:32.700 but maybe, maybe that's not a normal thing. 00:10:32.700 --> 00:10:34.340 You're such a C programmer. 00:10:34.340 --> 00:10:39.240 All right. 00:10:39.240 --> 00:10:40.820 Well, I think this is really cool. 00:10:40.820 --> 00:10:41.760 This is a specialist. 00:10:41.760 --> 00:10:47.040 And, you know, I don't know if I have any code that does math at that finer, greater level 00:10:47.040 --> 00:10:51.100 that I really care, but maybe, you know, if you're in charge of a library where you've 00:10:51.100 --> 00:10:55.720 got a tight loop or you do a lot of math science stuff where it matters, this can be really 00:10:55.720 --> 00:10:56.280 useful. 00:10:56.280 --> 00:11:02.420 And what's cool is it's not like, and switch to rust or switch to C or switch to Cython 00:11:02.420 --> 00:11:03.800 and it'll take effect. 00:11:03.800 --> 00:11:06.340 Like, no, this, this is like straight Python code. 00:11:06.340 --> 00:11:10.940 This is just, how do I take most advantage of what is already happening for performance 00:11:10.940 --> 00:11:13.200 boosts in 3.11 that we haven't had before? 00:11:13.200 --> 00:11:16.000 I think, and I think it's going to be just one more workflow step. 00:11:16.160 --> 00:11:22.220 So you've got, you profile your code, your whole thing is a little bit slower than you'd 00:11:22.220 --> 00:11:22.760 like it to be. 00:11:22.760 --> 00:11:24.440 You throw a profiler on it. 00:11:24.440 --> 00:11:27.200 You see the bottleneck areas that you could improve. 00:11:27.200 --> 00:11:32.640 And you think, should I like rewrite some of this in rust or C or, you know, what should 00:11:32.640 --> 00:11:32.980 I do? 00:11:32.980 --> 00:11:39.560 Well, first off, let's try doing this, like throw, throw, throw this at it and, and, and 00:11:39.560 --> 00:11:42.720 have the optimizer from 3.11 help you out. 00:11:42.720 --> 00:11:49.300 And, and yeah, so I think this, I can definitely see that this is going to be part of people's 00:11:49.300 --> 00:11:50.860 workflow, but yeah. 00:11:50.860 --> 00:11:52.500 I agree that you want to profile it. 00:11:52.500 --> 00:11:54.160 Yes, exactly. 00:11:54.160 --> 00:11:56.000 Cause while it's fun to do this. 00:11:56.000 --> 00:11:56.740 Yeah. 00:11:56.740 --> 00:11:57.340 Yeah. 00:11:57.340 --> 00:11:58.720 Only focus where it's going to matter. 00:11:58.720 --> 00:12:01.100 Don't, don't optimize a bunch of stuff that doesn't. 00:12:01.240 --> 00:12:06.540 So Brian out in the audience says, different Brian, is there a plan to do lossless type 00:12:06.540 --> 00:12:09.900 conversion or maybe flake eight can make this kind of suggestion? 00:12:09.900 --> 00:12:10.940 Yeah, exactly. 00:12:10.940 --> 00:12:12.100 That'd be fair. 00:12:12.100 --> 00:12:12.360 Yeah. 00:12:12.360 --> 00:12:18.020 I'm not really sure if you don't want to write the code where you get different outputs probably. 00:12:18.020 --> 00:12:18.620 Right. 00:12:18.620 --> 00:12:23.860 But everything that was happening here, you were, you ended up with the same outcome anyway. 00:12:23.860 --> 00:12:26.460 It's just like, well, do I do the division first or the multiplication? 00:12:27.000 --> 00:12:31.800 Or do I start with an ant that results after some addition subtraction with a float? 00:12:31.800 --> 00:12:33.520 Or is I just make them all floats? 00:12:33.520 --> 00:12:33.980 Right. 00:12:33.980 --> 00:12:38.280 I feel like it's, in most cases, it shouldn't be changing the outcome. 00:12:38.280 --> 00:12:38.880 So. 00:12:38.880 --> 00:12:39.520 Okay. 00:12:39.520 --> 00:12:40.220 Yeah. 00:12:40.220 --> 00:12:40.700 Yeah. 00:12:40.700 --> 00:12:41.240 Cool. 00:12:41.240 --> 00:12:43.740 Anyway, that's, that's what I got for the first one. 00:12:43.740 --> 00:12:44.120 How about you? 00:12:44.120 --> 00:12:48.800 Well, we're kind of sticking with a 3.11 theme so far. 00:12:48.800 --> 00:12:56.860 Well, we can use Toml now, but in 3.11, we are going to have a Toml that'd be part of 00:12:56.860 --> 00:13:00.040 part of Python 3.11 with PEP 6.80. 00:13:00.040 --> 00:13:02.700 And we covered that in episode 273. 00:13:02.700 --> 00:13:11.200 But I, I, one of the things we didn't mention was that, was the Toml lib is, is, 00:13:11.200 --> 00:13:15.800 and I think we did mention it, it's based on Tomly, but Tomly you can use right now. 00:13:15.800 --> 00:13:23.220 So a lot of projects are switching to use Tomly as a, as their Toml, Toml parser, to, to 00:13:23.220 --> 00:13:28.640 read, read, like pyproject.toml or, or read their own, config file. 00:13:28.980 --> 00:13:32.200 And, and so I just wanted to highlight it. 00:13:32.200 --> 00:13:36.060 It's a Tomly is the, a little Toml parser. 00:13:36.060 --> 00:13:39.300 it's a cute little thing on the project. 00:13:39.300 --> 00:13:39.820 It's cute. 00:13:39.820 --> 00:13:46.780 But, but I was reminded of it because, real, the real Python people, put 00:13:46.780 --> 00:13:49.120 out actually looks like gear. 00:13:49.500 --> 00:13:51.860 Sorry, I'm not going to try to pronounce that name. 00:13:51.860 --> 00:13:58.340 real Python, wrote an article called, Python and Toml new, new best friends. 00:13:58.340 --> 00:14:02.140 And I really love it's a, it's a very comprehensive article. 00:14:02.140 --> 00:14:07.660 but I really love at least the first three parts of it, using Toml as a config format, 00:14:07.660 --> 00:14:13.880 uh, getting to know key value pairs and, load Toml with Python, because this is kind 00:14:13.880 --> 00:14:15.140 of what you're going to do with it. 00:14:15.140 --> 00:14:16.780 You're going to write config files for something. 00:14:17.220 --> 00:14:21.680 And I just kind of, it's, this is a great introduction of Toml for Python. 00:14:21.680 --> 00:14:23.400 And that's kind of what we care about. 00:14:23.400 --> 00:14:23.660 Right. 00:14:23.660 --> 00:14:30.000 So, it goes through like just getting, getting used to what Toml looks like, what a 00:14:30.000 --> 00:14:35.200 config file looks like talking about how all the keys, even if you, it's like key value 00:14:35.200 --> 00:14:35.620 stuff. 00:14:35.620 --> 00:14:39.600 And even if you, you put a number there or something, it's going to be a string. 00:14:39.600 --> 00:14:42.640 All the keys get converted to strings, even if they don't look like them. 00:14:42.640 --> 00:14:46.460 and they are, they're, they're UTF eight. 00:14:46.460 --> 00:14:50.700 So you can use, unit code in there as well, which is kind of neat. 00:14:50.700 --> 00:14:52.360 put your emojis in there. 00:14:52.360 --> 00:14:53.020 Yeah. 00:14:53.140 --> 00:14:56.220 Well, can you, is, is our emojis UTF eight? 00:14:56.220 --> 00:15:00.880 I think mostly, many of them are interesting. 00:15:00.880 --> 00:15:05.320 That'd be fun to put, put emojis in your, I don't know. 00:15:05.320 --> 00:15:06.280 What motor we were in. 00:15:06.280 --> 00:15:08.520 Are we running in cow mode or lizard mode? 00:15:08.520 --> 00:15:09.200 I'll do lizard. 00:15:09.200 --> 00:15:09.440 Yeah. 00:15:09.440 --> 00:15:09.600 Okay. 00:15:09.600 --> 00:15:11.100 Well, if you're running in lizard mode. 00:15:11.100 --> 00:15:11.980 Okay. 00:15:11.980 --> 00:15:14.580 I got to try that to see. 00:15:14.580 --> 00:15:16.180 I should have done that before. 00:15:16.180 --> 00:15:16.760 Oh my gosh. 00:15:16.760 --> 00:15:21.960 I think almost it's both horrible and amazing to imagine writing like config files to like 00:15:21.960 --> 00:15:23.320 put it and put it in lizard mode. 00:15:23.320 --> 00:15:23.660 Do it. 00:15:23.660 --> 00:15:24.360 Yeah. 00:15:24.760 --> 00:15:28.040 one of the things that I didn't before reading this article, one of the things I 00:15:28.040 --> 00:15:32.820 didn't know you could do in Toml, because I just sort of cursory, I use it with pipe project 00:15:32.820 --> 00:15:40.420 at Toml and that's about it, but you can do, so, talks about, normal, how 00:15:40.420 --> 00:15:40.980 to read stuff. 00:15:40.980 --> 00:15:44.140 But one of the things is, Oh, what was I going to talk about? 00:15:44.140 --> 00:15:50.860 Arrays, and you can do arrays of things which are neat and tables and arrays of tables, 00:15:50.860 --> 00:15:56.160 uh, which is like, so you have arrays of tables or these bracket bracket things. 00:15:56.160 --> 00:15:59.300 And, and then you can do dot stuff. 00:15:59.300 --> 00:16:06.260 So if you have like, was it user and user dot player, these will show up as, 00:16:06.260 --> 00:16:09.760 as like, you know, sub dictionary key things. 00:16:10.180 --> 00:16:15.680 and so one of the things that I, and I played with it this morning and, it really, 00:16:15.680 --> 00:16:20.560 I should have had a something to show, but the thing I like to do is to just read it. 00:16:20.560 --> 00:16:26.100 just like, this article talks about reading it, just read the Toml file into Python 00:16:26.100 --> 00:16:27.040 and print it. 00:16:27.040 --> 00:16:33.000 and then you can, and it'll print out as a dictionary and then you can create whatever 00:16:33.000 --> 00:16:34.700 format you want for your Toml file. 00:16:34.700 --> 00:16:36.600 And then you can just see what it's going to look like. 00:16:36.600 --> 00:16:38.240 And then you know how to access it. 00:16:39.240 --> 00:16:40.980 That's one of the best ways to do that. 00:16:40.980 --> 00:16:41.600 That's awesome. 00:16:41.600 --> 00:16:42.460 Yeah. 00:16:42.460 --> 00:16:43.680 What an interesting format. 00:16:43.680 --> 00:16:45.060 That's pretty, that's pretty in depth. 00:16:45.060 --> 00:16:49.040 And, a blast from last week past Ashley. 00:16:49.040 --> 00:16:54.900 Hey, Ashley says UTF eight can encode any Unicode character emoji, your heart emoji. 00:16:54.900 --> 00:16:55.740 Heart out of it. 00:16:55.740 --> 00:16:56.000 Very. 00:16:56.000 --> 00:16:57.180 Oh yeah. 00:16:57.180 --> 00:16:59.460 You could do like, you know, is it in heart mode? 00:16:59.460 --> 00:17:00.380 Heart equals true. 00:17:00.380 --> 00:17:01.240 Heart equals false. 00:17:01.740 --> 00:17:06.540 oh, for optimize optimizer, you could do a flame emoji equals true. 00:17:06.540 --> 00:17:07.800 exactly. 00:17:07.800 --> 00:17:08.860 So I love it. 00:17:08.860 --> 00:17:09.260 Yeah. 00:17:09.260 --> 00:17:13.500 I think, look, we have not leveraged the configuration as emoji sufficiently. 00:17:13.500 --> 00:17:14.480 Oh yeah. 00:17:14.480 --> 00:17:18.920 I think, I think a pie test should rewrite all of its config figs as emoji items. 00:17:18.920 --> 00:17:19.760 Just do a PR. 00:17:19.760 --> 00:17:20.640 I'm sure we'll take it. 00:17:20.640 --> 00:17:20.900 Yeah. 00:17:20.900 --> 00:17:23.500 Yeah. 00:17:23.500 --> 00:17:23.900 It'd be good. 00:17:23.980 --> 00:17:24.260 All right. 00:17:24.260 --> 00:17:24.960 Yeah. 00:17:24.960 --> 00:17:25.300 All right. 00:17:25.300 --> 00:17:27.660 Let me tell you about our sponsor for this week before we move on. 00:17:27.660 --> 00:17:31.080 So this week is brought to you by Microsoft Founders Hub. 00:17:31.080 --> 00:17:34.440 In fact, they are supporting a whole bunch of upcoming episodes. 00:17:34.440 --> 00:17:37.240 So thank you a whole bunch to Microsoft for startups here. 00:17:37.240 --> 00:17:39.360 Starting business is hard by some estimates. 00:17:39.360 --> 00:17:43.140 Over 90% of startups go out of business within their first year. 00:17:43.140 --> 00:17:47.060 With that in mind, Microsoft for startups set out to understand what startups need to be 00:17:47.060 --> 00:17:51.000 successful and to create a digital platform to help overcome those challenges. 00:17:51.000 --> 00:17:53.340 Microsoft for startups Founders Hub. 00:17:53.340 --> 00:17:58.860 Their hub provides all founders at any stage with a bunch of free resources to help solve 00:17:58.860 --> 00:17:59.300 challenges. 00:17:59.300 --> 00:18:04.880 And you get technology benefits, but also really importantly, access to experts, guidance and 00:18:04.880 --> 00:18:08.320 skilled resources, mentorship and networking connections and a bunch more. 00:18:08.320 --> 00:18:14.360 So unlike a bunch of other similar projects in the industry, Microsoft for startup founders 00:18:14.360 --> 00:18:19.680 hub does not require startups to be investor backed or third party validated to participate. 00:18:19.680 --> 00:18:21.280 It's free to apply. 00:18:21.280 --> 00:18:22.960 And if you apply, get in. 00:18:23.180 --> 00:18:24.420 Then it's you're in. 00:18:24.420 --> 00:18:25.380 It's open to all. 00:18:25.380 --> 00:18:28.080 So what do you get if you join or apply and then get accepted? 00:18:28.080 --> 00:18:32.980 So you can speed up your development with access to GitHub, Microsoft cloud, the ability to unlock 00:18:32.980 --> 00:18:38.140 credits over time, as in it gets over a hundred thousand dollars worth of credits over time over 00:18:38.140 --> 00:18:38.680 the first year. 00:18:38.680 --> 00:18:41.080 If you meet a bunch of milestones, which is fantastic. 00:18:41.080 --> 00:18:42.760 Help your startup innovate. 00:18:42.760 --> 00:18:47.540 Founders Hub is partnering with companies like OpenAI, the global leader in AI research and 00:18:47.540 --> 00:18:49.760 development to provide benefits and discounts too. 00:18:50.160 --> 00:18:50.760 Yeah. 00:18:50.760 --> 00:18:54.720 Through Microsoft startup founders hub, becoming a founder is no longer about who you know. 00:18:54.720 --> 00:19:00.620 You'll have access to the mentorship network, giving you all access to a pool of hundreds of mentors across a range of disciplines. 00:19:01.160 --> 00:19:07.580 Areas like idea, validation, fundraising, management, coaching, sales marketing, as well as specific technical technical stress points. 00:19:07.580 --> 00:19:12.400 To me, that's actually the biggest value is the networking and mentor side. 00:19:12.400 --> 00:19:17.020 So you'll be able to book a one-on-one meeting with these mentors, many of whom are founders themselves. 00:19:17.020 --> 00:19:22.160 Make your idea a reality today with the critical support you'll get from Microsoft or startups founders hub. 00:19:22.160 --> 00:19:25.460 Join the program at pythonbytes.fm/founders hub. 00:19:25.760 --> 00:19:27.220 Link will be in your player's show notes. 00:19:27.220 --> 00:19:27.840 Nice. 00:19:27.840 --> 00:19:28.340 Yeah. 00:19:28.340 --> 00:19:28.740 Cool. 00:19:28.740 --> 00:19:29.620 Indeed. 00:19:29.620 --> 00:19:30.440 All right. 00:19:30.440 --> 00:19:32.160 I guess I'm up next with this order we got. 00:19:32.160 --> 00:19:35.460 And oh my goodness, Samuel Colvin, take a bow. 00:19:35.460 --> 00:19:42.060 Because he put out a plan for what's happening with Pydantic version two. 00:19:42.600 --> 00:19:52.760 But the reason I say take a bow is this is one detailed plan that is really, really thought through, thought out, backed up with a bunch of GitHub discussions and so on. 00:19:52.760 --> 00:19:53.300 Wow. 00:19:53.440 --> 00:19:57.560 So the idea is Pydantic started out as an interesting idea. 00:19:57.560 --> 00:20:04.300 And surprise, surprise, a bunch of people glommed onto it probably more than it was originally envisioned to be so. 00:20:04.300 --> 00:20:14.440 So, for example, SQL model from Sebastian Ramirez is like, Pydantic models are now our ORM to the database with all the interesting stuff that ORMs have. 00:20:14.440 --> 00:20:16.380 And Roman Wright said, guess what? 00:20:16.380 --> 00:20:18.860 We could do that for MongoDB as well. 00:20:18.860 --> 00:20:21.740 Same with the Pydastic thing we recently spoke about. 00:20:21.740 --> 00:20:28.740 And then Sebastian Ramirez is like, also, like, hey, FastAPI, this can be both our data exchange as well as our documentation. 00:20:28.740 --> 00:20:31.900 I was like, oh my goodness, what's going on here? 00:20:31.900 --> 00:20:39.740 So there's a bunch of stuff on the insides that could be better, let's say, or maybe time to rethink this. 00:20:39.740 --> 00:20:46.380 So in this plan, it talks about what they'll add, what they'll remove, what will change, some of the ideas for how long it will take, and so on. 00:20:46.380 --> 00:20:47.720 Yeah. 00:20:47.720 --> 00:20:49.660 Here's a pretty significant thing. 00:20:49.660 --> 00:20:55.660 I'm currently taking a kind of sabbatical after leaving my last job to work on this, which goes until October. 00:20:55.660 --> 00:21:00.800 So that's a big commitment to I'm going to help make Pydantic better. 00:21:00.800 --> 00:21:01.540 Oh, wow. 00:21:01.540 --> 00:21:02.960 It sounds familiar. 00:21:02.960 --> 00:21:07.140 It sounds a bit like Rich and Textual and those types of things as well. 00:21:07.220 --> 00:21:11.920 But this is a big, big commitment from Samuel, and he's really doing a ton of work. 00:21:11.920 --> 00:21:14.120 It says, people seem to care about my project. 00:21:14.120 --> 00:21:17.380 It's downloaded 26 million times a month. 00:21:17.380 --> 00:21:18.440 Wow. 00:21:18.440 --> 00:21:19.400 Which is insane. 00:21:19.400 --> 00:21:20.440 Yeah, it's awesome. 00:21:20.440 --> 00:21:22.140 That's kind of incredible. 00:21:22.140 --> 00:21:22.920 It is. 00:21:22.920 --> 00:21:24.940 And so it says, here's the basic roadmap. 00:21:24.940 --> 00:21:28.740 Implement a few features in what's now called the Pydantic core. 00:21:28.860 --> 00:21:31.880 We just had Ashley, who, as we saw, is out in the audience. 00:21:31.880 --> 00:21:32.380 Hey, Ashley. 00:21:32.380 --> 00:21:34.500 Who give a bit of a shout out to this feature. 00:21:34.500 --> 00:21:39.500 And also, I do want to also credit a couple other people's because Douglas Nichols and John 00:21:39.500 --> 00:21:42.240 Thagan also let me know that this was big news coming. 00:21:42.240 --> 00:21:43.740 So thank you all for that. 00:21:43.740 --> 00:21:49.480 The Pydantic core is being rewritten in Rust, which doesn't mean you have to know or do anything. 00:21:49.480 --> 00:21:51.520 It just means you have to pip install something. 00:21:51.520 --> 00:21:54.440 You get a binary compiled thing that runs a lot faster. 00:21:54.820 --> 00:21:56.040 Okay, so more on that in a second. 00:21:56.040 --> 00:22:01.980 First, they're working to get 110 out and basically merge every open PR that makes sense 00:22:01.980 --> 00:22:04.940 and close every PR that doesn't make sense. 00:22:04.940 --> 00:22:09.820 And then profusely apologize to why your PR that you've spent a long time making was closed 00:22:09.820 --> 00:22:10.380 without merging. 00:22:10.380 --> 00:22:12.860 Some other bookkeeping things. 00:22:12.860 --> 00:22:17.000 Start tearing the Pydantic code apart and see how many existing tests can still be made 00:22:17.000 --> 00:22:19.300 to pass and then release eventually Pydantic. 00:22:19.300 --> 00:22:22.560 The goal is to have this done by October, probably by the end of the year for sure. 00:22:22.560 --> 00:22:24.200 A couple of things worth paying attention to. 00:22:24.320 --> 00:22:26.040 There are a bunch of breaking changes in here. 00:22:26.040 --> 00:22:31.500 A lot of things are being cleaned up, reorganized, renamed, some removed. 00:22:31.500 --> 00:22:34.800 Like from ORM, people might be using that with SQLAlchemy. 00:22:34.800 --> 00:22:37.220 That's being removed, for example, and so on. 00:22:37.220 --> 00:22:43.100 So there's, if you depend heavily on Pydantic, especially if you build a project like Beanie 00:22:43.100 --> 00:22:46.700 that depends heavily on Pydantic, you're going to need to look at this because some of the 00:22:46.700 --> 00:22:47.520 stuff won't work anymore. 00:22:47.520 --> 00:22:49.660 But let's highlight a couple of things here. 00:22:49.660 --> 00:22:50.100 Performance. 00:22:50.660 --> 00:22:57.080 This one is really important because this is the data exchange level for FastAPI. 00:22:57.080 --> 00:23:00.040 This is the database transformation level. 00:23:00.040 --> 00:23:03.760 When I do a query from the database, what comes back comes back in some raw form and then is 00:23:03.760 --> 00:23:05.000 turned into a Pydantic model. 00:23:05.380 --> 00:23:09.380 And those are computationally expensive things that happen often. 00:23:09.380 --> 00:23:17.420 And in general, Pydantic version 2 is about 17 times 1,700% faster than V1 when validating 00:23:17.420 --> 00:23:19.700 models in a standard scenario. 00:23:19.700 --> 00:23:22.900 It says between 4 to 50 times faster than Pydantic 1. 00:23:22.900 --> 00:23:23.400 Hmm. 00:23:23.400 --> 00:23:23.940 Wow. 00:23:24.040 --> 00:23:24.600 That's cool, right? 00:23:24.600 --> 00:23:25.240 Yeah. 00:23:25.240 --> 00:23:30.680 That alone should make your ears perk up and go, excuse me, my ORM just got 17 times faster. 00:23:30.680 --> 00:23:31.580 Wait a minute. 00:23:31.580 --> 00:23:32.620 I'm liking this. 00:23:32.620 --> 00:23:37.840 I know that this is not the only thing that happens at ORM level, but the ones that, the 00:23:37.840 --> 00:23:41.960 ones I called out that depend heavily on it, like that's in the transformation path. 00:23:41.960 --> 00:23:42.680 So this is important. 00:23:42.680 --> 00:23:43.100 Yeah. 00:23:43.100 --> 00:23:45.920 This is actually, I'm super impressed. 00:23:46.040 --> 00:23:51.780 I have not, I normally don't even see this sort of advanced planning in commercial projects. 00:23:51.780 --> 00:23:52.720 Yes. 00:23:52.720 --> 00:23:53.320 Oh yeah. 00:23:53.320 --> 00:23:57.400 You could do a whole business startup that doesn't have the amount of thought that went 00:23:57.400 --> 00:24:00.160 into like what's happening in the next version of Pydantic. 00:24:00.160 --> 00:24:00.880 It's ridiculous. 00:24:00.880 --> 00:24:01.740 Yeah. 00:24:01.740 --> 00:24:02.840 It's incredible. 00:24:02.840 --> 00:24:04.540 I mean, I was serious when I said take a bow. 00:24:04.540 --> 00:24:09.740 It really lays out, opens a discussion about certain things and so on. 00:24:09.740 --> 00:24:11.140 So like another one is strict mode. 00:24:11.140 --> 00:24:14.400 I think I even saw a comment in the chat about it. 00:24:14.740 --> 00:24:19.440 So one of the things I actually like about Pydantic, but under certain circumstances, 00:24:19.440 --> 00:24:23.500 I can see why you would not want it is if you have something you say is an integer field 00:24:23.500 --> 00:24:26.120 and then you pass one, two, three, the number rate. 00:24:26.120 --> 00:24:31.020 But if you also pass quote one, two, three, Pydantic will magically parse that for you. 00:24:31.020 --> 00:24:32.640 Like this happens all the time on the internet. 00:24:32.640 --> 00:24:35.580 Like a query string has a number, but query strings are always strings. 00:24:35.580 --> 00:24:37.740 There's no way to have anything but strings. 00:24:37.740 --> 00:24:38.240 Yeah. 00:24:38.240 --> 00:24:39.320 So you got to convert them. 00:24:39.320 --> 00:24:39.500 Right. 00:24:39.500 --> 00:24:41.240 So this automatically does that. 00:24:41.240 --> 00:24:44.420 But if you don't want that to happen, you say, you gave me a string. 00:24:44.440 --> 00:24:45.080 It's invalid. 00:24:45.080 --> 00:24:48.000 You can turn on strict mode, which is off by default, I believe. 00:24:48.000 --> 00:24:49.340 There's also a bunch of plain. 00:24:49.340 --> 00:24:49.820 Go ahead. 00:24:49.820 --> 00:24:53.140 So strict mode does the conversion or strict mode? 00:24:53.140 --> 00:24:54.760 Strict mode won't do the conversion. 00:24:54.760 --> 00:24:56.420 It says, you said it's an int. 00:24:56.420 --> 00:24:57.140 You gave me a string. 00:24:57.140 --> 00:24:57.960 Nope. 00:24:57.960 --> 00:25:00.140 Rather than, could it be an integer? 00:25:00.140 --> 00:25:01.580 Let's try that first. 00:25:01.580 --> 00:25:02.380 You know what I mean? 00:25:02.380 --> 00:25:03.080 Yeah. 00:25:03.140 --> 00:25:11.700 Maybe one of the things you do is, in the ORM level, one of those things, you might put it in strict mode so it doesn't do as much work trying to convert stuff. 00:25:11.700 --> 00:25:12.880 I don't know if that actually would matter. 00:25:12.880 --> 00:25:14.780 But formalizes a bunch of conversions. 00:25:15.300 --> 00:25:18.760 It has built-in JSON support and different things. 00:25:18.760 --> 00:25:26.960 Another big thing is this Pydantic core will be able to be used outside of Pydantic classes now. 00:25:27.400 --> 00:25:39.260 So you can do a significant performance to improve stuff like adding validation to data classes or validating arguments and query strings or a type dick or a function argument or whatever. 00:25:39.260 --> 00:25:39.860 Yeah. 00:25:39.860 --> 00:25:40.600 Yeah. 00:25:40.600 --> 00:25:41.680 Let's see. 00:25:41.680 --> 00:25:42.660 Next up. 00:25:42.660 --> 00:25:44.480 And let's see. 00:25:44.480 --> 00:25:45.340 This one. 00:25:45.340 --> 00:25:46.080 Strict mode. 00:25:46.080 --> 00:25:47.360 We talked about strict mode. 00:25:47.360 --> 00:25:50.040 Another one is required versus nullable. 00:25:50.040 --> 00:25:56.400 There's a little bit of ambiguity of, you know, if you said something's a string, that means it's required and it can't be none. 00:25:56.740 --> 00:26:06.180 If you say it's a string type none or it's an optional string or something like that, then basically the behaviors were a little bit different. 00:26:06.180 --> 00:26:09.980 So originally, I think this is when typing was pretty new. 00:26:09.980 --> 00:26:14.460 It said Pydantic previously had a confused idea of required versus nullable. 00:26:14.460 --> 00:26:24.680 This mostly resulted from Sam's misgivings about marking a field as optional but requiring a value to be provided to it but allowing it to be set to none or something along those lines. 00:26:25.680 --> 00:26:28.460 Anyway, there's minor changes around that. 00:26:28.460 --> 00:26:29.440 Let's see. 00:26:29.440 --> 00:26:32.060 Final one that I want to cover is namespace stuff. 00:26:32.060 --> 00:26:35.740 And this is like a whole bunch of things are now getting renamed. 00:26:36.000 --> 00:26:43.380 So for example, if you override, if you implemented or overrode validate JSON, it's now model underscore validate JSON. 00:26:43.380 --> 00:26:45.900 If you had is instance, it's now model is instance. 00:26:45.900 --> 00:26:46.520 Okay. 00:26:46.520 --> 00:26:48.380 There's a bunch of these changes all over the place. 00:26:48.380 --> 00:26:48.920 Yeah. 00:26:48.920 --> 00:26:51.140 That look like they're going to cause breaking changes. 00:26:51.140 --> 00:26:51.940 They're easy to fix. 00:26:51.940 --> 00:26:52.720 Just change the name. 00:26:52.720 --> 00:26:54.300 But, you know, it's not nothing. 00:26:54.740 --> 00:26:56.180 Also, parse file. 00:26:56.180 --> 00:26:59.040 I still love his hander here. 00:26:59.040 --> 00:27:00.040 Pars file. 00:27:00.040 --> 00:27:01.200 This was a mistake. 00:27:01.200 --> 00:27:02.700 It should have never been in Pydantic. 00:27:02.700 --> 00:27:03.640 We're removing it. 00:27:03.640 --> 00:27:04.100 Okay. 00:27:04.100 --> 00:27:04.500 Pars raw. 00:27:04.500 --> 00:27:07.680 Partially replaced by this other thing. 00:27:07.680 --> 00:27:09.160 Anything else it did was a mistake. 00:27:09.160 --> 00:27:11.680 From ORM, this has been moved somewhere else. 00:27:11.680 --> 00:27:13.040 Schema and so on. 00:27:13.040 --> 00:27:15.380 So you just, like, there's a lot of stuff that people are using here. 00:27:15.380 --> 00:27:16.800 So just have a look. 00:27:16.800 --> 00:27:17.520 Try it out. 00:27:17.520 --> 00:27:19.260 Don't just go, oh, then version 2 is out. 00:27:19.260 --> 00:27:19.900 Is this going to work? 00:27:19.980 --> 00:27:21.780 Like, this is going to have some significant changes. 00:27:21.780 --> 00:27:27.400 And another reason why it's really awesome that he goes through so much detail is because 00:27:27.400 --> 00:27:29.960 there's going to be stuff that breaks. 00:27:29.960 --> 00:27:32.920 So it's a breaking interface change. 00:27:32.920 --> 00:27:36.400 And so, yeah, it's cool that it's this detailed. 00:27:36.400 --> 00:27:38.360 And a couple things to notice. 00:27:38.360 --> 00:27:40.480 Let's see. 00:27:40.480 --> 00:27:42.680 Somebody else in the chat mentioned. 00:27:42.680 --> 00:27:44.200 Richard mentioned. 00:27:44.200 --> 00:27:46.380 And he has emojis in the headers. 00:27:46.380 --> 00:27:48.300 Yeah, there's emojis in the headers. 00:27:48.500 --> 00:27:54.740 And I got to say, like, the navigation in the table of contents, very cool. 00:27:54.740 --> 00:27:59.040 It goes to, like, light gray for areas you've already seen. 00:27:59.040 --> 00:27:59.480 And then. 00:27:59.480 --> 00:28:00.620 Oh, that's interesting. 00:28:00.620 --> 00:28:01.760 It's a cool thing. 00:28:01.760 --> 00:28:03.460 Yeah, it's quite cool. 00:28:03.460 --> 00:28:05.340 I think it went on and on. 00:28:05.340 --> 00:28:06.460 But two real quick things. 00:28:06.460 --> 00:28:10.640 One, there'll be no pure Python implementation of the core. 00:28:10.640 --> 00:28:11.560 It's always Rust. 00:28:11.560 --> 00:28:15.200 But they list out the platforms where it'll be compiled to, including WebAssembly. 00:28:15.200 --> 00:28:16.260 Oh, nice. 00:28:16.780 --> 00:28:22.080 They previously had some Cython in what was supposed to be pure Python's Pydantic. 00:28:22.080 --> 00:28:30.660 And so now a kind of bonus is the Pydantic model, the Pydantic package, becomes a pure Python package, whereas previously it wasn't. 00:28:30.660 --> 00:28:35.420 So they've taken, like, all of that behavior and put it under this core thing that ships as a Rust binary. 00:28:35.420 --> 00:28:40.000 And now instead of doing some Cython middle ground, it's pure Python again. 00:28:40.000 --> 00:28:42.460 So that's interesting refactoring, I think. 00:28:42.460 --> 00:28:43.060 Yeah. 00:28:43.060 --> 00:28:43.740 Yeah. 00:28:43.740 --> 00:28:43.940 Yeah. 00:28:43.940 --> 00:28:45.540 And finally, documentation. 00:28:45.540 --> 00:28:52.020 When you get a validation error, it gives you a link to the documentation in the JSON error message. 00:28:52.020 --> 00:28:53.260 That's pretty cool. 00:28:53.260 --> 00:28:54.260 Yeah. 00:28:54.260 --> 00:28:55.100 That's nice. 00:28:55.100 --> 00:28:56.040 All right. 00:28:56.220 --> 00:28:56.440 Yeah. 00:28:56.440 --> 00:28:58.340 Anyway, that's quite a plan, isn't it, Brian? 00:28:58.340 --> 00:28:58.940 Yeah. 00:28:58.940 --> 00:28:59.820 Quite a plan. 00:28:59.820 --> 00:29:00.300 All right. 00:29:00.300 --> 00:29:01.940 Well, I'm excited for it. 00:29:01.940 --> 00:29:03.280 Okay. 00:29:03.280 --> 00:29:07.680 Well, next topic is a little more lighthearted. 00:29:07.680 --> 00:29:09.200 It's about fish. 00:29:09.200 --> 00:29:12.540 Pike, to be specific. 00:29:12.540 --> 00:29:14.920 No, it's about PDFs. 00:29:15.140 --> 00:29:19.560 So it's just a cool project I noticed. 00:29:19.560 --> 00:29:20.820 Pike PDF. 00:29:20.820 --> 00:29:24.180 It's a Python library for reading and writing PDF files. 00:29:24.180 --> 00:29:25.280 What's the big deal? 00:29:25.280 --> 00:29:26.340 We've had these before. 00:29:26.340 --> 00:29:33.060 But this is, it's based on QPDF, which is a C++-based library. 00:29:33.060 --> 00:29:38.000 And it's presently continued being maintained. 00:29:38.000 --> 00:29:41.060 So it's kind of pretty fast. 00:29:41.060 --> 00:29:44.540 Well, actually, I'm assuming it's fast if it's C++ in the background. 00:29:44.540 --> 00:29:44.580 Yeah. 00:29:44.580 --> 00:29:51.980 But it's also pretty just nice and elegant to do things. 00:29:51.980 --> 00:29:56.020 And the documentation has this nice fish, which is good. 00:29:56.020 --> 00:29:59.460 I always like cool diagrams, cool logos. 00:29:59.460 --> 00:30:03.740 But some of the neat things that you can do with it. 00:30:03.740 --> 00:30:08.480 So it's recommending that you not use it if you're just writing PDF files. 00:30:08.480 --> 00:30:11.980 That there's other things that you can use. 00:30:12.120 --> 00:30:12.440 What was it? 00:30:12.440 --> 00:30:14.680 Like Report Lab to write PDFs. 00:30:14.680 --> 00:30:19.300 But if you're having to read or modify PDFs, then this is where it shines. 00:30:19.300 --> 00:30:22.780 You can do things like copy pages from one PDF to another. 00:30:22.780 --> 00:30:24.800 Split and merge PDFs. 00:30:24.800 --> 00:30:27.500 Extract content out of PDFs. 00:30:27.500 --> 00:30:30.920 Like if you're using it for data stuff. 00:30:30.920 --> 00:30:34.880 You get a report in PDF and you're trying to pull the information out. 00:30:34.880 --> 00:30:37.000 You can use it for that. 00:30:37.000 --> 00:30:37.580 Or images. 00:30:37.680 --> 00:30:39.600 You can pull all the images out of a PDF file. 00:30:39.600 --> 00:30:41.120 Or this is kind of cool. 00:30:41.120 --> 00:30:47.220 You can replace images in a PDF file and generate a new one without changing anything else about the file. 00:30:47.220 --> 00:30:48.900 It's kind of neat. 00:30:48.900 --> 00:30:56.760 So just kind of a neat, if people are working with reading or modifying PDF files, maybe check this one out. 00:30:57.020 --> 00:30:57.860 Yeah, this looks great. 00:30:57.860 --> 00:31:01.980 The fact that it's in C++, I'm guessing it's probably standalone. 00:31:01.980 --> 00:31:08.540 I remember I've done some PDF things before and it felt like I had to install some OS level thing that it shelled out to. 00:31:08.540 --> 00:31:09.280 So this is cool. 00:31:09.620 --> 00:31:19.820 Nice on the ReadMe, it has a comparison of some of the different PDF libraries that you could use. 00:31:19.820 --> 00:31:24.320 And some of the reasons why you might want this one, like it supports more versions. 00:31:24.320 --> 00:31:31.720 I didn't realize that one of these libraries I've heard of before, PDF-RW, doesn't support the newer versions. 00:31:31.720 --> 00:31:32.700 So bummer. 00:31:33.400 --> 00:31:39.100 And then also password-protected files, it supports that. 00:31:39.100 --> 00:31:42.900 Except for, but not public key ones, but just normal passwords. 00:31:42.900 --> 00:31:44.120 Straight passwords, yeah. 00:31:44.120 --> 00:31:44.480 Yeah. 00:31:44.480 --> 00:31:44.900 That's great. 00:31:44.900 --> 00:31:45.720 So it's kind of neat. 00:31:45.720 --> 00:31:52.420 Also like the measure of actively maintained, the commit activity per year over the year or something like that. 00:31:52.420 --> 00:31:53.120 Oh, right. 00:31:53.120 --> 00:31:54.220 That's kind of interesting. 00:31:54.220 --> 00:31:55.680 Yeah, it's an interesting metric. 00:31:55.680 --> 00:31:56.620 It seems good. 00:31:56.620 --> 00:31:58.240 I haven't really thought about it lately, but. 00:31:58.240 --> 00:31:58.680 Yeah. 00:31:58.680 --> 00:31:59.220 Yeah. 00:31:59.220 --> 00:31:59.800 Nice. 00:31:59.800 --> 00:32:01.140 All right, yeah, this is a great one. 00:32:01.740 --> 00:32:04.700 Well, so that's it for our main items. 00:32:04.700 --> 00:32:06.220 Yeah, what else you got? 00:32:06.220 --> 00:32:07.100 Any extras? 00:32:07.100 --> 00:32:13.100 Well, last week, we talked about the critical packages. 00:32:13.100 --> 00:32:14.060 Critical packages. 00:32:14.060 --> 00:32:16.080 Or at some recent. 00:32:16.080 --> 00:32:20.320 Yeah, last week, we talked about critical packages. 00:32:20.320 --> 00:32:23.440 Either yesterday or last week, depending on how you consume this material. 00:32:23.440 --> 00:32:23.500 Exactly. 00:32:23.500 --> 00:32:24.480 Yeah. 00:32:24.800 --> 00:32:31.440 So I was surprised to find out that pytestCheck, the plugin I wrote, was one of those. 00:32:31.440 --> 00:32:32.260 I'm like, really? 00:32:32.260 --> 00:32:34.620 Because it's like the top 1%. 00:32:34.620 --> 00:32:39.180 So if anybody's curious, I wanted to just highlight that a little bit. 00:32:39.180 --> 00:32:43.200 So pytestCheck is a plugin that allows multiple failures per test. 00:32:44.020 --> 00:32:49.920 And one of the best ways, it's a secondary way that one of the contributors added, is you 00:32:49.920 --> 00:32:51.380 can use it as a context manager. 00:32:51.380 --> 00:32:54.100 You can say, like, with check, and then do an assert. 00:32:54.100 --> 00:32:56.260 And then you're going to have multiple of those within a. 00:32:56.260 --> 00:32:57.640 I like the one-liner even. 00:32:57.640 --> 00:32:58.720 That's pretty nice. 00:32:58.720 --> 00:32:59.080 Yeah. 00:32:59.080 --> 00:33:03.460 And this is totally, like, black will totally reformat this if you ran it through black. 00:33:03.460 --> 00:33:04.820 But it's nice. 00:33:04.820 --> 00:33:06.120 You'd have to block it out. 00:33:06.240 --> 00:33:09.500 Anyway, I was like, how could it be? 00:33:09.500 --> 00:33:13.120 Well, I'm curious what on the list it was. 00:33:13.120 --> 00:33:18.940 So there's a place called, what, HugoVK. 00:33:18.940 --> 00:33:22.360 Has a top PyPI packages list. 00:33:22.360 --> 00:33:23.320 And it's updated. 00:33:23.320 --> 00:33:25.260 I think it's just updated once a month or something. 00:33:25.260 --> 00:33:27.260 But you can do the top 5. 00:33:27.260 --> 00:33:29.440 You can do the top 5,000. 00:33:29.440 --> 00:33:32.500 Yeah, it's the top 5,000 or 1,000 or 100. 00:33:33.440 --> 00:33:38.020 And so I was curious about where on the list I was. 00:33:38.020 --> 00:33:41.140 I'm number 1,677. 00:33:41.140 --> 00:33:43.140 So kind of far down the list. 00:33:43.140 --> 00:33:45.220 But hey, we're just talking. 00:33:45.220 --> 00:33:47.940 It's still in the top third of the top 1%. 00:33:47.940 --> 00:33:48.680 That's pretty awesome. 00:33:48.680 --> 00:33:51.380 The pytest is number 72. 00:33:51.380 --> 00:33:52.560 That was pretty neat. 00:33:52.560 --> 00:33:57.440 And Pydantic, which we covered, was, I just checked, 117. 00:33:57.440 --> 00:34:03.100 But there are 57 pytest plugins that show up in the top 3,500. 00:34:03.520 --> 00:34:04.400 So that's pretty neat. 00:34:04.400 --> 00:34:04.520 Wow. 00:34:04.520 --> 00:34:05.000 Pretty neat. 00:34:05.000 --> 00:34:05.960 That is pretty neat. 00:34:05.960 --> 00:34:07.620 That's all I got for extras. 00:34:07.620 --> 00:34:08.640 All right. 00:34:08.640 --> 00:34:09.920 Well, I have zero extras. 00:34:09.920 --> 00:34:11.420 So mine are finished as well. 00:34:11.420 --> 00:34:12.340 How about a joke? 00:34:12.340 --> 00:34:12.960 Yeah. 00:34:12.960 --> 00:34:13.500 Great. 00:34:13.500 --> 00:34:14.000 All right. 00:34:14.000 --> 00:34:15.460 I told you we're coming back to it. 00:34:15.460 --> 00:34:17.100 So this one comes from Netta. 00:34:17.100 --> 00:34:21.740 Netta Code Girl at Netta.mk. 00:34:21.740 --> 00:34:24.260 And let me just pull this one up here. 00:34:24.260 --> 00:34:24.440 Right. 00:34:24.500 --> 00:34:28.640 So this one is, there's this colleague here. 00:34:28.640 --> 00:34:29.240 Can I make this? 00:34:29.240 --> 00:34:29.620 There we go. 00:34:29.620 --> 00:34:30.080 Make it a little bigger. 00:34:30.080 --> 00:34:37.160 There's the two women who are developers, Netta and her unnamed friend who always has gotten 00:34:37.160 --> 00:34:38.980 in trouble with the elevator last time, basically. 00:34:39.320 --> 00:34:44.620 And there's this sort of weird manager looking guy that comes in and says, I tested your 00:34:44.620 --> 00:34:47.540 chat bot, but some of its replies are really messed up. 00:34:47.540 --> 00:34:50.040 Well, that's what testing is all about. 00:34:50.040 --> 00:34:52.620 I'll go through the logs later, says one of the girls. 00:34:52.620 --> 00:34:53.380 No, no, no. 00:34:53.380 --> 00:34:54.160 No, no, no. 00:34:54.160 --> 00:34:54.660 No, no. 00:34:54.660 --> 00:34:55.160 No need. 00:34:55.160 --> 00:34:56.520 Right. 00:34:56.520 --> 00:34:57.500 Check out the faces. 00:34:57.500 --> 00:34:59.160 She's like, excuse me. 00:34:59.160 --> 00:35:02.020 I'm not even sure I want to open the logs now. 00:35:02.020 --> 00:35:02.580 Yeah. 00:35:02.580 --> 00:35:02.920 Yeah. 00:35:02.920 --> 00:35:04.160 Don't look at the logs. 00:35:04.880 --> 00:35:06.000 That's what testing's for. 00:35:06.000 --> 00:35:08.660 I'll go through the logs later. 00:35:08.660 --> 00:35:13.940 Well, yeah, she's got some good ones in her list there. 00:35:13.940 --> 00:35:14.900 So love it. 00:35:14.900 --> 00:35:15.340 Yeah. 00:35:15.340 --> 00:35:16.680 I like the art too. 00:35:16.680 --> 00:35:17.940 Nice art. 00:35:17.940 --> 00:35:18.360 I do too. 00:35:18.360 --> 00:35:19.020 It is. 00:35:19.020 --> 00:35:21.120 So also nice was our podcast. 00:35:21.120 --> 00:35:22.020 Thanks for being here. 00:35:22.020 --> 00:35:22.780 Thank you. 00:35:22.780 --> 00:35:23.300 Yeah. 00:35:23.300 --> 00:35:23.680 You bet. 00:35:23.680 --> 00:35:24.700 See you next week.