Trascrizione completa di #561: TonIO, a Multi-threaded Async Runtime for Python

idea. 00:04:26.740 --> 00:04:27.500 I want to put it out there. 00:04:27.800 --> 00:04:29.520 If it gains no traction, you can shut it down. 00:04:30.080 --> 00:04:32.000 If it gains a ton of traction, you're excited. 00:04:32.100 --> 00:04:33.280 If it gains a little traction, 00:04:33.780 --> 00:04:35.140 but enough, then you got to keep working on it. 00:04:35.200 --> 00:04:36.740 That's kind of not ideal, right? 00:04:37.080 --> 00:04:37.320 Yeah. 00:04:37.430 --> 00:04:39.060 And especially like nowadays with AI, 00:04:39.560 --> 00:04:41.440 like the amount of things you, 00:04:41.570 --> 00:04:42.520 oh, I have an idea. 00:04:43.740 --> 00:04:48.680 Like the chances like to just throw a bunch of agents into that and see what happens. 00:04:49.200 --> 00:04:49.840 It's very dangerous. 00:04:50.240 --> 00:04:50.820 It's very dangerous. 00:04:51.240 --> 00:04:52.040 I have an idea. 00:04:52.040 --> 00:04:52.840 It'll probably take me an hour. 00:04:52.880 --> 00:04:54.080 Let me see if I can validate it. 00:04:54.520 --> 00:04:57.080 Thursday afternoon, you're still like, oh, I'm still almost there. 00:04:57.120 --> 00:04:58.980 It's like, where did my week go? 00:04:59.280 --> 00:04:59.500 What happened? 00:05:00.360 --> 00:05:01.420 It's fun though, isn't it? 00:05:01.500 --> 00:05:02.360 It's a wild time. 00:05:02.720 --> 00:05:05.000 Yeah, it's weird, right? 00:05:05.200 --> 00:05:12.800 Because in a way, we have this very powerful, not super constant tool. 00:05:14.080 --> 00:05:19.760 but it's very powerful and and sometimes just amazing like sometimes like what do you what 00:05:19.760 --> 00:05:27.160 you get back is just amazing and you say like oh okay that's nice that's unexpected sometimes is 00:05:27.280 --> 00:05:34.960 the dumbest thing possible you can you you can get but i guess like the point the main point is that 00:05:35.240 --> 00:05:42.860 i think it it shows up how as humans we have like a complicated relationship with tools 00:05:43.860 --> 00:05:51.980 it can exploit a bunch of weird human behavior sometimes so yeah yeah if you like building things 00:05:52.520 --> 00:05:56.560 you know create you've got ideas and you want to see them come to life it's an unprecedented time 00:05:56.720 --> 00:06:03.220 if if you really your main joy was to be in the code working on the little bits of nuance i can 00:06:03.300 --> 00:06:07.300 see that it's frustrating and uh there's there's a whole there's a whole thing that we go into there 00:06:07.260 --> 00:06:08.120 but that's probably not. 00:06:08.130 --> 00:06:09.660 But wow, what a crazy time, right? 00:06:09.900 --> 00:06:10.900 Yeah, absolutely. 00:06:11.400 --> 00:06:14.000 Yeah, so let's jump into the topics here. 00:06:14.010 --> 00:06:18.560 I think the first thing I want to do is let's just do a little dive into Granian. 00:06:19.320 --> 00:06:21.520 That's why I reached out to you to have you on the show the first time. 00:06:21.840 --> 00:06:23.160 Like I said, it's powering. 00:06:23.390 --> 00:06:25.460 I have on the Talk Python server, 00:06:25.680 --> 00:06:29.680 there's 33 different containers running a bunch of different things, 00:06:29.770 --> 00:06:32.440 and probably about 15 of them are running Granian. 00:06:33.020 --> 00:06:33.380 Nice. 00:06:33.880 --> 00:06:34.160 Yeah. 00:06:34.800 --> 00:06:36.820 So like I said, it's been really, really cool. 00:06:37.040 --> 00:06:41.400 just tell people what is Granian because I think not only is it interesting as a thing that you've 00:06:41.400 --> 00:06:45.580 done previously and been on the show before it's also an interesting lead-in to where we're going 00:06:45.580 --> 00:06:52.940 with Tonio so what's great yeah sure uh so Granian is an application server for Python so in Python 00:06:53.540 --> 00:07:02.700 for historical reasons um we separated like the application layer from the HTTP layer so when we 00:07:02.600 --> 00:07:07.640 want to serve a web application in Python compared to other languages 00:07:07.740 --> 00:07:12.260 in which the server part is actually a part of the application. 00:07:12.640 --> 00:07:18.640 In Python, we tend to have a separated package, which is a server, 00:07:19.100 --> 00:07:22.340 and that server's import your actual application. 00:07:22.760 --> 00:07:25.820 And then we have application protocols 00:07:26.720 --> 00:07:30.120 to make the server talk with your application and back. 00:07:31.100 --> 00:07:34.960 The two main protocols in Python are VUSGI or WSGI. 00:07:35.500 --> 00:07:38.120 I never really understood how to say that, 00:07:39.720 --> 00:07:44.760 which is like the original PET333, if I'm not mistaken. 00:07:45.050 --> 00:07:48.100 So it's quite ancient in the Python history. 00:07:49.140 --> 00:07:53.860 And once we had like AsyncIO, a new protocol emerged. 00:07:54.380 --> 00:07:57.660 That's called ASGI or ASGI because, of course, 00:07:57.780 --> 00:08:00.080 it's the asynchronous server gateway. 00:08:00.520 --> 00:08:05.260 So, Granian is, again, an application server, 00:08:06.360 --> 00:08:11.720 so you can consider it as an alternative to MicroVoosgi, 00:08:13.860 --> 00:08:17.600 Uvicorn, Hypercorn, and a bunch of other servers 00:08:17.820 --> 00:08:19.240 that happened during the years. 00:08:20.520 --> 00:08:24.640 And, yeah, the peculiar fact is it's made with Rust. 00:08:26.080 --> 00:08:29.880 So, compared to all the other servers you might use in Python, 00:08:29.960 --> 00:08:34.659 all the HTTP part or the network layer 00:08:35.300 --> 00:08:39.780 is running completely out of the interpreter, 00:08:40.940 --> 00:08:46.020 which kind of alleviates the load on the interpreter, 00:08:46.230 --> 00:08:47.320 like on the Python side. 00:08:48.550 --> 00:08:50.680 And depending on the context, 00:08:50.860 --> 00:08:53.420 it might be more performant 00:08:53.660 --> 00:08:57.500 or it might stabilize your latency more. 00:08:58.660 --> 00:09:03.360 You know, I think

thinking back now, I'm pretty sure that that was the reason I chose Granian. 00:09:03.770 --> 00:09:09.420 Not that it was a ton faster, but the P95 bad side of things. 00:09:09.550 --> 00:09:13.260 Like how much might it slow down under certain weird circumstances? 00:09:13.760 --> 00:09:17.800 It was way more stable than a lot of the other servers, right? 00:09:17.870 --> 00:09:19.220 And that's kind of what you're referring to. 00:09:19.820 --> 00:09:23.640 Yeah, because again, like a normal server which runs everything into Python, 00:09:23.720 --> 00:09:39.480 Like it has this kind of side effects that when your application is starting becoming slow, it kind of becomes like an infinite loop in which like the server adds additional load to the interpreter, which is already like highly loaded from your application. 00:09:39.860 --> 00:09:49.940 And so like Graniac kind of avoids all of that because like even if your application is fully loaded, like all the web and network layer is out of the interpreter. 00:09:50.100 --> 00:09:52.880 It doesn't disturb, let's say the interpreter with that. 00:09:53.200 --> 00:09:53.280 Yeah. 00:09:53.540 --> 00:09:58.080 I do want to just give a little bit of a hat tip to WSGI or WSGI. 00:09:58.450 --> 00:10:14.240 Because, yeah, you pointed out it is old, but it's really, really nice that if somebody is out there running on GUnicorn or they're running on some other server, MicroWSGI, which you should stop running MicroWSGI because it's not supported anymore. 00:10:14.600 --> 00:10:16.580 You don't have to change your code at all. 00:10:16.760 --> 00:10:20.480 You just take your flask or your Jenga or whatever, and you just say, now run here. 00:10:20.510 --> 00:10:24.740 And they just, Grandin talks WizGey or Asggy to that thing, and it's transparent. 00:10:25.330 --> 00:10:25.720 It's beautiful. 00:10:26.140 --> 00:10:26.240 Yeah. 00:10:26.510 --> 00:10:26.620 Yeah. 00:10:26.940 --> 00:10:33.659 I mean, there are like still some, you know, sometimes weird things like, like, Vertzoic 00:10:34.500 --> 00:10:37.680 has some subtleties, sometimes some nuances. 00:10:37.840 --> 00:10:50.700 So I had to like add some environment key just for Verzoic because under some circumstances, if that key is not there, things don't work exactly good. 00:10:51.120 --> 00:11:02.120 So yes, I agree that like having like the two protocols separated and coded like in a specific way gives you this ability to switch the server. 00:11:02.520 --> 00:11:08.180 On the other hand, sometimes, you know, you can be a pain in some time. 00:11:08.680 --> 00:11:09.400 Yeah, for sure. 00:11:09.840 --> 00:11:10.460 Yeah, for sure. 00:11:11.780 --> 00:11:18.380 But yeah, I'd say like we like recently it's getting more popular. 00:11:18.690 --> 00:11:27.540 So I'd say if I'm not mistaken, like a few weeks ago, I think I surpassed like HyperCorn in terms of downloads. 00:11:28.760 --> 00:11:28.980 Awesome. 00:11:29.060 --> 00:11:37.480 And yeah,

it's used like in a bunch of different big companies, Microsoft, Google, for sure. 00:11:38.350 --> 00:11:40.000 We use it at Sentry, of course. 00:11:41.140 --> 00:11:46.040 And I think like Sentry is the only company in which we use like all the three supported 00:11:46.380 --> 00:11:51.300 protocols of Granian because Granian also supports like its own protocol, which is named 00:11:51.410 --> 00:11:56.320 like RSGI, I don't know, like RSGI, which is a redesign. 00:11:56.740 --> 00:11:56.940 Risky. 00:11:57.840 --> 00:11:57.940 Yeah. 00:12:00.220 --> 00:12:07.460 um so yeah i'd say it's going pretty good um i think like the last let's say this year was mostly 00:12:07.800 --> 00:12:16.340 into um not not not not mostly about features i think features but mostly about you know making 00:12:16.480 --> 00:12:22.400 the thing more stable and covering like some edge cases um especially again we're using it the 00:12:22.460 --> 00:12:28.260 Sentry. So now like if something is wrong with Grinia, like I have people knocking on my door. 00:12:28.900 --> 00:12:34.900 I bet you do. Well, I mean, with Sentry, Sentry gets an insane amount of like an unimaginable 00:12:35.040 --> 00:12:39.480 amount of traffic. I'm sure with the error reporting side of things and probably a little 00:12:39.580 --> 00:12:46.040 more so these days on the MCP and API side as well with all the agents. I know that I use the MCP 00:12:46.120 --> 00:12:50.580 and absolutely love it. It's been incredible. It's really, really good. I could just say on my 00:12:50.600 --> 00:12:54.720 project hey there's a century error i got what's up like literally that's all i gotta say in clause 00:12:54.840 --> 00:12:57.920 like hold on we're on it and it's like yeah you know five minutes later it's like here's what's 00:12:57.930 --> 00:13:02.580 going on let's work on it you know yeah really good stuff uh i do want to give a quick shout out as 00:13:02.700 --> 00:13:08.220 well uh on this regard to gradient like people are like oh it's it's not as popular as server x or 00:13:08.310 --> 00:13:12.120 whatever right although i think that's starting to fade it's it's got 5 000 stars and a ton of 00:13:12.210 --> 00:13:19.079 people using it and so on yeah but it's also based on hyper from rust which itself has uh where where's 00:13:19.100 --> 00:13:25.420 my numbers, 410,000 projects using it. So it's really tested, right? And obviously the stuff at 00:13:25.500 --> 00:13:31.640 Sentry as well. Yeah. Yeah. Like the only thing I'm sure about Granian is that like the HTTP stack, 00:13:32.100 --> 00:13:38.280 that's super safe. Like that's like, I could bet like everything I have on that. So if something 00:13:38.300 --> 00:13:44.660 is wrong in Granian, like that's on me, not on Hyper, like to be clear. This portion of Talk 00:13:44.660 --> 00:13:49.660 Python to me is brought to you by Sentry and Sear AI. There are plenty of AI tools that help you 00:13:50.080 --> 00:13:55.280 write code, but Sentry Sear is built to help you fix it when it breaks. The difference is context. 00:13:55.920 --> 00:14:00.500 Sear isn't just guessing based on syntax. It's analyzing your actual Sentry data, 00:14:00.760 --> 00:14:05.360 your stack traces, logs, and failure patterns. Because it has the full context, it can 00:14:05.740 --> 00:14:09.620 A, spot buggy code in review and help prevent issues before they happen, 00:14:10.260 --> 00:14:13.040 and B, identify the root cause of production errors. 00:14:13.980 --> 00:14:18.540 It can even draft a fix and hand the work off to an agent-like cursor to open a PR for you. 00:14:19.140 --> 00:14:21.140 Seer turns Sentry into a complete loop. 00:14:21.660 --> 00:14:24.500 You have your traces, errors, logs, and replays to see the problem, 00:14:24.800 --> 00:14:26.260 and now AI to help solve it. 00:14:26.620 --> 00:14:30.920 Join millions of devs at companies like Claude, Disney+, and even Talk Python, 00:14:31.380 --> 00:14:32.560 who use Sentry to move faster. 00:14:33.080 --> 00:14:37.860 Check them out at talkpython.fm/sentry and use code talkpython26, 00:14:38.440 --> 00:14:41.800 all one word, for $100 in Sentry credits. 00:14:42.440 --> 00:14:44.180 Thank you to Sentry for supporting Talk Python. 00:14:45.660 --> 00:14:48.980 I recently did a post over on my personal blog 00:14:49.080 --> 00:14:52.860 calling cutting Python web app memory by 31%. 00:14:52.860 --> 00:14:55.760 And I did a ton of analysis and stuff 00:14:56.660 --> 00:14:59.260 on how I was running Granian and different things, 00:14:59.300 --> 00:15:00.240 just how I was running Python. 00:15:00.820 --> 00:15:02.400 And I think this is going to be a good lead-in 00:15:02.400 --> 00:15:03.540 to where we're going as well. 00:15:03.880 --> 00:15:13.040 I got it to where I was using 1.3 gigs, let's say, for Talk Python training the courses specifically. 00:15:13.360 --> 00:15:15.160 And people are like, oh, it's just a little courses thing. 00:15:15.200 --> 00:15:19.960 Like, well, it has 178,000 lines of Python, so it's not a totally small project. 00:15:20.440 --> 00:15:23.240 But down by the end, I did a bunch of stuff on it. 00:15:24.480 --> 00:15:25.280 Lazy imports. 00:15:25.779 --> 00:15:28.660 I switched to async, which is why I brought this up. 00:15:28.940 --> 00:15:31.160 But somewhere down here, I don't know what the final number is. 00:15:31.400 --> 00:15:31.820 Here we go. 00:15:31.880 --> 00:15:36.560 In the end, I got it down to 450 megs from 0.28 gigs. 00:15:37.190 --> 00:15:40.900 One of the main reasons I did that as a way I want to kind of want to lead into like this, 00:15:41.380 --> 00:15:45.740 where we're going is I said, well, the way grading works is for each, 00:15:46.250 --> 00:15:49.540 either you can make multi-processes that are single threaded or it could do multiple threads 00:15:49.660 --> 00:15:51.380 and then it'll create a separate interpreter for each one. 00:15:51.490 --> 00:15:55.240 Either way, it

kind of like replicates out the running. 00:15:55.420 --> 00:15:58.660 It'll create copies in one way or another of your code. 00:15:59.010 --> 00:16:01.760 And that, if you could run one of them, that would be great. 00:16:01.840 --> 00:16:03.000 But why do we have so many? 00:16:03.330 --> 00:16:04.900 We have so many because of the GIL. 00:16:05.010 --> 00:16:06.140 We have poor concurrency. 00:16:06.640 --> 00:16:10.360 It's really hard to handle a lot of load and like actually access, you know, my servers, 00:16:10.600 --> 00:16:10.920 eight cores. 00:16:11.010 --> 00:16:14.060 Like I could only access one core like normally, right? 00:16:14.380 --> 00:16:17.560 And so we do all these things to like explode them out. 00:16:17.680 --> 00:16:18.880 But what is the most expensive? 00:16:19.140 --> 00:16:22.160 Like five years ago, this was true on servers, but really is true now. 00:16:22.250 --> 00:16:23.000 What is most expensive? 00:16:23.360 --> 00:16:24.380 Memory, not compute, right? 00:16:25.180 --> 00:16:25.300 Yeah. 00:16:25.700 --> 00:16:25.760 Yeah. 00:16:26.160 --> 00:16:26.600 Thanks AI. 00:16:27.000 --> 00:16:27.540 Thanks AI. 00:16:27.710 --> 00:16:27.940 Thanks Altman. 00:16:28.860 --> 00:16:30.120 We really appreciate that. 00:16:32.260 --> 00:16:34.580 But we've free-threaded Python now. 00:16:34.780 --> 00:16:35.300 That's a thing. 00:16:35.740 --> 00:16:43.280 And so we're kind of moving into this world where we can take better advantage of the hardware and we don't have to multiply out. 00:16:44.220 --> 00:16:47.060 I know your process takes 500 megs of RAM, but we need four of them. 00:16:47.090 --> 00:16:49.580 So now you need two gigs just to run your code. 00:16:50.360 --> 00:16:51.940 We don't necessarily need that anymore. 00:16:52.160 --> 00:16:53.780 And grading would be a really good foundation for that. 00:16:54.000 --> 00:16:59.440 But let's just think broadly about free-threading and async as it is and stuff. 00:16:59.660 --> 00:17:02.620 I know you've thought a lot about this because of this project that you built. 00:17:03.680 --> 00:17:04.819 Well, what do you think? 00:17:05.350 --> 00:17:08.720 First thing, can I say free threading to me? 00:17:09.140 --> 00:17:11.199 It's super exciting. 00:17:11.640 --> 00:17:16.060 And I don't know why people are not as much excited as me on this. 00:17:16.819 --> 00:17:25.480 Because this is, I'd say, the biggest thing that ever happened into Python since the beginning of Python. 00:17:25.680 --> 00:17:27.140 In the last 30 years. 00:17:27.380 --> 00:17:31.720 do you remember like Python 2 to Python 3 and all like, 00:17:32.080 --> 00:17:36.360 oh, now strings are not bytes anymore and everything is Unicode? 00:17:36.700 --> 00:17:36.800 Yeah. 00:17:37.040 --> 00:17:40.960 That's nothing compared to, hey, the gil is not there anymore. 00:17:41.080 --> 00:17:46.800 Like we talk about like every Python room everywhere in the world, 00:17:46.860 --> 00:17:53.220 like at some point talked about the gil and why we have to do to have the gil, right? 00:17:53.960 --> 00:17:54.160 Right. 00:17:54.720 --> 00:17:59.220 The Gilectomy that Larry Hastings had, it seemed to perpetually be working on and never 00:17:59.440 --> 00:18:00.020 kind of went anywhere. 00:18:00.340 --> 00:18:00.480 Yeah. 00:18:00.480 --> 00:18:05.660 It took like 15 years of people talking about this and it finally landed. 00:18:05.860 --> 00:18:12.000 Like last year, yes, of course, last year it was beta because like in Python 3.13, it 00:18:12.000 --> 00:18:14.380 was like, this is an experiment still. 00:18:14.900 --> 00:18:17.840 That was the asterisk on top. 00:18:18.960 --> 00:18:22.120 But since Python 3.14, that's considered stable. 00:18:22.540 --> 00:18:28.560 It's exactly as the normal GIL-included interpreter. 00:18:30.480 --> 00:18:33.540 And at the end of the day, the biggest part of all of this 00:18:33.800 --> 00:18:38.800 was using reference counting instead of singleton. 00:18:39.160 --> 00:18:44.000 So I guess I'm excited about free-threaded Python 00:18:44.020 --> 00:18:47.840 because finally I have the language I love the most. 00:18:49.020 --> 00:18:53.180 not because of the structure, not because of types or how it's designed, 00:18:53.180 --> 00:18:57.840 but because of this awesome community that only Python language has, 00:18:58.220 --> 00:18:59.080 at least in my experience. 00:19:00.220 --> 00:19:04.080 And so finally, I can use this language with actual threads. 00:19:04.540 --> 00:19:08.460 So I'm not obliged anymore to think, oh, wait, we are in Python, 00:19:08.920 --> 00:19:13.760 so I need to use a C library or something to take advantage of my CPU, 00:19:14.480 --> 00:19:25.040 Especially given, I don't know, we are in the era of, hey, even if you buy a $200 CPU that probably has 12 cores or something. 00:19:26.380 --> 00:19:28.420 So yeah, I'm very excited. 00:19:29.700 --> 00:19:32.740 I'm confused about the lack of adoption. 00:19:33.160 --> 00:19:35.260 I would love to see more adoption in this. 00:19:35.500 --> 00:19:46.120 And of course, like there are reasons, like every native C compiled extension has to be fixed to be thread safe. 00:19:47.640 --> 00:19:53.360 And probably like on a bunch of ancient extensions, that's not simple or easy. 00:19:55.180 --> 00:19:56.240 Yeah, I don't know, man. 00:19:56.320 --> 00:19:58.100 Like I'm just too excited. 00:19:58.380 --> 00:20:00.620 Like I want people to use free thread. 00:20:00.800 --> 00:20:02.020 I know. 00:20:02.180 --> 00:20:04.260 And getting it is so much easier as well. 00:20:04.340 --> 00:20:10.840 it's just uv Python install, you know, that 3.14 T and you got it, right? 00:20:10.980 --> 00:20:11.200 Yeah. 00:20:11.600 --> 00:20:11.720 Yeah. 00:20:12.080 --> 00:20:16.740 And like, I don't know, I'm so pissed at the maintainer of the Docker images 00:20:16.880 --> 00:20:20.240 because we don't have like an official 3.14 free threaded image.

00:20:20.360 --> 00:20:20.660 Like why? 00:20:21.160 --> 00:20:21.220 Why? 00:20:21.400 --> 00:20:23.460 Oh yeah, that is actually a good point. 00:20:23.580 --> 00:20:28.900 Like it has been a year and a half and it just requires putting the T on the uv installed bit. 00:20:29.160 --> 00:20:29.440 Yeah. 00:20:29.740 --> 00:20:29.860 Yeah. 00:20:30.320 --> 00:20:30.460 Yeah. 00:20:30.480 --> 00:20:36.660 Yeah, but again, to my perspective, free-threaded Python opens the window to do a bunch of... 00:20:36.840 --> 00:20:42.040 First, to solve a bunch of the issues you mentioned, which is, for example, in Centri, 00:20:42.160 --> 00:20:43.980 the number one issue is... 00:20:44.120 --> 00:20:50.460 I mean, Centri at the end of the day is this 18-years-old Django monolith. 00:20:51.420 --> 00:20:59.520 And so, yes, we deploy in several multiple different deployments, activating just some parts. 00:20:59.720 --> 00:21:20.080 But even so, like, yeah, we have the same problem, which is, hey, like, in order to use like one C view out of like a VM that GCP gives us, like, yeah, we need to spawn like four different processes and require like, I don't know, four, five, six gigabytes of memory for like a single pod. 00:21:20.320 --> 00:21:31.320 And not only do you have to do that, you also have to, anytime there's some sort of in-memory cache, like an LRU cache or something, every one of those has to be spun up for all four, not just for one, right? There's a lot of powerful sharing. 00:21:31.440 --> 00:21:34.120 Do we want to start talking about observability? 00:21:34.420 --> 00:21:36.580 Because you cannot just use like 00:21:36.640 --> 00:21:39.380 the Prometheus SDK or anything like that. 00:21:40.000 --> 00:21:42.000 Because again, like then those became like 00:21:42.480 --> 00:21:43.340 four different instances. 00:21:43.600 --> 00:21:45.800 So you cannot even track like the same metrics 00:21:46.620 --> 00:21:48.480 unless you do like some weird tricks. 00:21:48.960 --> 00:21:53.200 Like, yeah, relying on the multiprocessing library. 00:21:54.180 --> 00:21:54.800 It's messy, man. 00:21:54.980 --> 00:21:57.080 So like, yeah, I think like 00:21:57.620 --> 00:22:00.220 FreeThreader like solves a bunch of problems, 00:22:00.820 --> 00:22:04.180 Especially in web development in Python. 00:22:05.420 --> 00:22:09.780 Because everybody says like, oh yeah, they did this just because of AI, you know, 00:22:09.880 --> 00:22:12.240 because they can like make things parallel, whatever. 00:22:12.720 --> 00:22:14.200 I don't get that. 00:22:14.300 --> 00:22:18.280 Like, like we have like this amazing feature and it solves a bunch of problem 00:22:18.460 --> 00:22:23.180 on web development and also like it opens up the a bunch of new opportunities. 00:22:23.400 --> 00:22:27.680 Because again, like one of the most annoying parts in Granian, which is also 00:22:27.700 --> 00:22:33.400 Like one of the things that people opens like the most, the vast majority of issues about 00:22:33.580 --> 00:22:38.960 is because nobody understands like how socket sharing works across different processes. 00:22:39.370 --> 00:22:45.100 And again, like with FreeThreader, like that's just go away if we have something different 00:22:45.230 --> 00:22:45.940 from a Synco. 00:22:46.310 --> 00:22:51.280 And I guess that's the, that's when we get on the point here. 00:22:52.520 --> 00:22:57.660 Because for example, like in Grainium, so Grainium supports FreeThreader Python scenes 00:22:57.680 --> 00:22:59.440 2.0, if I'm not mistaken. 00:22:59.920 --> 00:23:01.520 And so when you use 00:23:01.700 --> 00:23:03.420 Granian on a free-threaded interpreter, 00:23:03.840 --> 00:23:05.520 workers become 00:23:07.600 --> 00:23:07.660 threads. 00:23:08.200 --> 00:23:09.340 They're no longer processes. 00:23:10.200 --> 00:23:11.520 And I can skip all the dance 00:23:11.620 --> 00:23:13.160 of, yeah, open the socket, 00:23:14.140 --> 00:23:15.500 but just bind 00:23:15.500 --> 00:23:17.440 on that, but not on listen, because then 00:23:17.640 --> 00:23:19.440 it gets to be shared across processes 00:23:19.960 --> 00:23:21.500 and so each worker can have his own 00:23:21.660 --> 00:23:23.540 backlog. It's a mess, man. 00:23:23.740 --> 00:23:25.700 So on the free-threaded variant, 00:23:26.200 --> 00:23:28.100 Like, Grenian already, like, used threads. 00:23:29.799 --> 00:23:35.280 The subtlety is that each thread has its own event loop. 00:23:35.510 --> 00:23:41.280 So what Grenian cannot do today is sharing things between those threads. 00:23:41.560 --> 00:23:49.180 At the end of the day, yes, the application is shared, but, like, every request, like, if a repus comes to a worker, that's it. 00:23:49.250 --> 00:23:50.120 Like, there's no... 00:23:50.310 --> 00:23:51.780 You cannot cross boundaries, right? 00:23:52.060 --> 00:23:54.040 when you await for something, 00:23:54.130 --> 00:23:55.260 when you suspend for something, 00:23:56.440 --> 00:23:57.500 that's the loop, 00:23:57.750 --> 00:23:59.020 the event loop you're running into. 00:23:59.980 --> 00:24:01.540 Which is a big limitation for servers. 00:24:02.300 --> 00:24:03.320 Yeah, and one is calling await 00:24:03.540 --> 00:24:04.400 and another is calling await 00:24:04.680 --> 00:24:06.060 or maybe doing some work, 00:24:06.260 --> 00:24:07.860 but if they were the same event loop, 00:24:08.220 --> 00:24:09.180 one thing could run 00:24:09.420 --> 00:24:10.240 because the other is awaiting, 00:24:10.380 --> 00:24:12.400 but they're actually separate altogether, right? 00:24:12.560 --> 00:24:15.460 Yeah, and also you can still end up 00:24:15.460 --> 00:24:16.720 in these weird conditions 00:24:16.930 --> 00:24:19.360 in which maybe, I don't know, 00:24:19.440 --> 00:24:21.139 50% of your connection 00:24:21.740 --> 00:24:24.640 we're bound to a specific event loop. 00:24:25.180 --> 00:24:28.080 And so you also have like unbalancement 00:24:28.190 --> 00:24:29.420 between like

the different threads. 00:24:29.510 --> 00:24:30.640 So you have like one thread 00:24:30.800 --> 00:24:32.880 that receives the vast majority of things, 00:24:33.060 --> 00:24:34.100 the other thread doing nothing. 00:24:35.480 --> 00:24:35.640 Yeah. 00:24:35.860 --> 00:24:36.000 Yeah. 00:24:36.390 --> 00:24:38.380 Let's say like we have new opportunities. 00:24:39.200 --> 00:24:41.080 We kind of lack some libraries, 00:24:41.370 --> 00:24:42.700 like some libraries behavior 00:24:43.700 --> 00:24:45.620 that can actually give us like 00:24:45.880 --> 00:24:47.580 all the actual meat here. 00:24:48.600 --> 00:24:48.900 Agreed. 00:24:49.020 --> 00:24:53.600 I want to talk about one area before we dive in, the Tone.io, 00:24:53.820 --> 00:24:55.740 and that's just asyncio event loops. 00:24:56.120 --> 00:24:57.460 And I know during your presentation, 00:24:57.700 --> 00:25:00.240 so you gave a talk on this at EuroPython this year, 00:25:00.340 --> 00:25:04.120 and I'll link to the topic right here in the moment. 00:25:05.140 --> 00:25:06.920 But there are people like, oh, well, we have uvloops, 00:25:07.040 --> 00:25:08.300 so doesn't that kind of solve it? 00:25:08.500 --> 00:25:10.980 And I think that's a really interesting thought 00:25:11.180 --> 00:25:14.080 because without calling out anybody, I didn't intend to call out anybody. 00:25:14.240 --> 00:25:18.540 I think it just highlights in the Python space, 00:25:18.940 --> 00:25:22.240 There's just not a lot of thinking about what concurrency means, 00:25:22.550 --> 00:25:24.660 why it matters, how you program for it. 00:25:25.460 --> 00:25:26.760 And that's going to be a challenge. 00:25:27.020 --> 00:25:28.620 I think that's part of the challenge of what you're talking about 00:25:28.660 --> 00:25:30.180 with the adoption of free-threaded Python. 00:25:30.330 --> 00:25:34.380 But I think also it is a challenge in the sense that 00:25:34.720 --> 00:25:37.520 thinking about making the event loop faster by, I don't know, 00:25:37.720 --> 00:25:39.900 20% or whatever uvloop actually does, which is great, 00:25:40.100 --> 00:25:43.300 it doesn't really address the fact that you're still only on one thread. 00:25:43.800 --> 00:25:47.400 Your event loop can access, like, I just, against probably better judgment, 00:25:47.600 --> 00:25:51.120 I just bought a Mac Studio Max, which has 18 cores. 00:25:51.559 --> 00:25:55.700 If I write one thread, that's about like 5% of the capacity of that machine. 00:25:56.300 --> 00:26:01.600 So what does it matter if it's 4.9% or 5.1% depending on which loop? 00:26:01.840 --> 00:26:02.520 I don't care. 00:26:02.640 --> 00:26:04.120 That is broken either way. 00:26:04.280 --> 00:26:06.400 I want 95% capacity. 00:26:06.780 --> 00:26:07.080 Yes. 00:26:07.280 --> 00:26:07.560 Right? 00:26:07.940 --> 00:26:08.920 And that's, I don't know. 00:26:08.960 --> 00:26:11.940 I thought it would be worthwhile to kind of like talk about uvloop a little bit. 00:26:12.020 --> 00:26:12.940 And you built our loop. 00:26:13.100 --> 00:26:14.860 And those things are great in their building blocks. 00:26:14.960 --> 00:26:15.100 Yeah. 00:26:15.180 --> 00:26:16.820 They're not the panacea. 00:26:16.920 --> 00:26:17.440 They're not the fix. 00:26:17.860 --> 00:26:25.900 Yeah, and we also now have Marcelo Trzesitsky, the maintainer of UVCore and Starlet. 00:26:27.680 --> 00:26:39.500 It has recently published Zoovloop, which is another event loop for Python relying on libUV, but it's written in Zig. 00:26:40.900 --> 00:26:44.280 But yeah, as you said, all of these event loops, 00:26:46.220 --> 00:26:52.580 yes, they can optimize some of the hot part of the cycle. 00:26:53.760 --> 00:26:55.320 But basically, that's it. 00:26:55.820 --> 00:27:00.960 Nothing really changed the shape of how things work. 00:27:02.080 --> 00:27:07.939 And so, yes, for sure, you should actually install 00:27:08.100 --> 00:27:13.780 uvloop or r loop or zoo v loop or whatever to to speed up like because that's performance you're 00:27:13.820 --> 00:27:19.320 leaving on the table regardless right like so uh it's good to have this project because they also 00:27:19.480 --> 00:27:26.500 like can light uh some of the inefficiencies of a sink io in my opinion um so i think it's still 00:27:26.550 --> 00:27:31.960 good to have these projects and you should use them um but yeah they want like the best you can 00:27:31.920 --> 00:27:39.520 GAT is probably like in the order of like 20% in, and we're talking about like row throughput 00:27:39.940 --> 00:27:46.180 if you're just doing TCP and nothing else, which is nobody does like, like nothing does 00:27:46.380 --> 00:27:47.020 nothing out there. 00:27:47.180 --> 00:27:47.240 Right? 00:27:47.320 --> 00:27:48.800 Like we do a bunch of stuff. 00:27:49.160 --> 00:27:49.400 Yeah. 00:27:49.480 --> 00:27:52.980 If you construct an example where basically all you're doing is waiting very efficiently, 00:27:53.410 --> 00:27:55.180 then you start to see those, those deviate. 00:27:55.180 --> 00:27:59.460 And also if you've got, like, I've got 10 tasks that run over two seconds. 00:27:59.940 --> 00:28:00.660 It does zero. 00:28:00.690 --> 00:28:01.220 It doesn't matter. 00:28:01.600 --> 00:28:05.080 This only matters when you're doing very, very fine-grained work and tons of it, 00:28:05.080 --> 00:28:07.900 and there's lots of switching and that sort of thing, right? 00:28:08.160 --> 00:28:09.080 Yes, yes. 00:28:11.120 --> 00:28:16.140 And so, yes, I guess this was one of my realizations. 00:28:16.570 --> 00:28:20.300 Because, again, I worked on ArtLoop mainly last year. 00:28:21.470 --> 00:28:24.900 And working on ArtLoop made me realize two things. 00:28:25.860 --> 00:28:31.260 The first thing is that asimkayo, it's overcomplicated sometimes. 00:28:31.540 --> 00:28:39.160 It's so fun fact at the presentation I did about that loop in Python Italy 25. 00:28:40.450 --> 00:28:51.900 I had a slide asking to the audience,

what's the difference between a protocol and a transport in Async.io. 00:28:52.220 --> 00:29:01.300 Because if you check the official documentation of Python and you get to that chapter, you will find out that they have like to try to explain you that difference. 00:29:01.860 --> 00:29:05.280 they try to do that like in three different ways, 00:29:06.340 --> 00:29:07.720 which to my experience, 00:29:07.880 --> 00:29:10.320 like if you need three different ways 00:29:10.380 --> 00:29:11.620 to explain the same thing, 00:29:12.020 --> 00:29:13.800 something is deeply wrong. 00:29:14.620 --> 00:29:17.860 Definitely violates an Einstein core philosophy. 00:29:18.180 --> 00:29:18.360 Yes. 00:29:19.260 --> 00:29:20.400 Also because like, again, 00:29:20.660 --> 00:29:22.460 if we think about the Zen of Python, 00:29:23.140 --> 00:29:25.280 like simple is better than complex, right? 00:29:27.000 --> 00:29:29.440 So that was my first realization, right? 00:29:30.440 --> 00:29:35.480 Because, again, like you have a bunch of primitives, like you have events, futures. 00:29:37.400 --> 00:29:43.880 Oh, by the way, every time they get spawned into the event loop, they get wrapped into a task. 00:29:44.200 --> 00:29:45.780 What is a task? Nobody knows. 00:29:46.920 --> 00:29:51.220 Oh, and by the way, when they actually run inside the event loop, they became handles. 00:29:52.380 --> 00:29:59.820 So we have a bunch of this weird thing that they're not really exposed to you if you just write async code. 00:30:00.180 --> 00:30:06.020 But if you need to debug something when something is wrong, then you need all of that to understand what's going on, right? 00:30:06.800 --> 00:30:14.120 And the other realization, again, was, okay, I reached a point in which there's no other possible optimization in our book. 00:30:14.300 --> 00:30:20.620 Because the vast majority of CPU time spent here is around AsyncIO primitives. 00:30:21.000 --> 00:30:22.560 And I have one single thread. 00:30:22.960 --> 00:30:28.920 And so, like, the only, when I came to realize this, I just said, okay, wait a second. 00:30:29.100 --> 00:30:35.020 Is that, is really that hard to start like scratch whiteboard? 00:30:35.420 --> 00:30:40.380 Like from the beginning, like let's just assume for a second, like asyncio never existed. 00:30:40.720 --> 00:30:42.920 And I want to do like async stuff in Python. 00:30:43.160 --> 00:30:43.860 What that look like? 00:30:43.980 --> 00:30:47.300 Like what, what does it mean to make like an asynchronous runtime in Python? 00:30:47.540 --> 00:30:48.100 Is that hard? 00:30:48.420 --> 00:30:55.119 And I don't know, like, I guess three or four weeks after this, I just got like an asynchronous 00:30:55.140 --> 00:30:57.040 runtime, a complete alternative 00:30:57.240 --> 00:30:59.040 asynchronous runtime running on Python with 00:30:59.220 --> 00:30:59.700 multiple threads. 00:31:01.280 --> 00:31:03.080 To be clear on free-threaded Python, right? 00:31:03.320 -->

00:31:05.160 Yes, free-threaded Python only because 00:31:05.650 --> 00:31:07.160 again, to have like real threads 00:31:07.530 --> 00:31:09.260 I mean, the design is to have 00:31:09.400 --> 00:31:11.120 real threads, so the moment you have 00:31:11.120 --> 00:31:12.860 the gill, I say, no, no, no, 00:31:13.200 --> 00:31:15.060 runtime error, change your 00:31:15.280 --> 00:31:17.200 interpreter because this is not good, 00:31:17.380 --> 00:31:17.520 right? 00:31:19.200 --> 00:31:21.080 This portion of Talk Python I May is brought to you by 00:31:21.300 --> 00:31:23.040 our AI tools. You know that 00:31:23.180 --> 00:31:25.100 thing where you ask an AI something about 00:31:25.120 --> 00:31:30.860 Python and it confidently tells you about a library version from 18 months ago? Well, we fixed that, 00:31:31.100 --> 00:31:37.480 at least for our shows. Talk Python and Python Bytes both have MCP servers now. Connect Talk Python 00:31:37.530 --> 00:31:43.780 and your AI can search over 550 episodes, full transcripts, every guest in the entire course 00:31:44.120 --> 00:31:49.820 catalog. Connect Python Bytes and you get almost 500 episodes of Python news going back to 2026, 00:31:50.520 --> 00:31:54.940 including every link we've ever put in the show notes. That means you can say things like, 00:31:55.060 --> 00:32:02.460 ask Talk Python what astral joining OpenAI means for uv or what has Python bytes said about uv 00:32:02.860 --> 00:32:06.740 and get real answers with real links, not hallucination. Name the show in your prompt 00:32:06.920 --> 00:32:12.900 and your AI knows exactly where to look. And if you live in the terminal, Talk Python also has a CLI 00:32:12.930 --> 00:32:20.900 too. One line, uv tool install talk-python-cli and then search episodes, transcripts, guests, 00:32:21.220 --> 00:32:26.880 courses without ever opening a browser. It's open source and it outputs text, JSON, and Markdown, 00:32:26.960 --> 00:32:31.980 so it feeds the AI tools that don't speak MCP yet. And here's the real reason I built it. 00:32:32.460 --> 00:32:38.320 Both shows cover around 10 years of Python history. The people, the decisions, the packages that took 00:32:38.380 --> 00:32:45.460 over, and the ones that quietly didn't. This enhanced access is free. No account, no API key, 00:32:45.560 --> 00:32:52.140 nothing to buy. This history should belong to all of us. Visit talkpython.fm and Python Bytes and 00:32:52.260 --> 00:32:56.840 click the MCP link in the nav bar. Connect them right now to your agents so that they'll be 00:32:57.080 --> 00:33:02.440 accessible anytime you need them in the future. And so yeah, that was like kind of the landscape 00:33:02.980 --> 00:33:09.180 in which I started building Tonio or Tonio. You can pronounce it like however you want. Like I'm 00:33:09.300 --> 00:33:13.540 Italian, so I say Tonio, of course. I think obviously, well, you have to do Tonio. 00:33:14.780 --> 00:33:15.980 So it's really good. 00:33:16.140 --> 00:33:21.900 But as it pairs to asyncio, tone IO also kind of like as a hat to that, right? 00:33:22.180 --> 00:33:22.300 Yeah. 00:33:22.560 --> 00:33:22.660 Yeah. 00:33:23.280 --> 00:33:24.040 But we have Trio. 00:33:24.460 --> 00:33:26.760 So that's not Trio. 00:33:28.020 --> 00:33:30.280 Trio is living there a little bit as well. 00:33:30.660 --> 00:33:30.740 Yeah. 00:33:30.800 --> 00:33:33.700 So this is a really interesting project that you have. 00:33:33.880 --> 00:33:37.140 And it basically says, what if we actually had threads? 00:33:37.560 --> 00:33:44.000 One of the things that's endlessly frustrated me about Python and asyncio and consequently 00:33:44.020 --> 00:33:48.220 I think the language implementation of async await is it's really good. 00:33:48.320 --> 00:33:49.200 I think, you know, 00:33:49.840 --> 00:33:52.940 the way it sort of turns async code into what looks like structurally 00:33:54.299 --> 00:33:57.060 serializable code or serial code is really, really nice. 00:33:57.320 --> 00:33:58.160 But the, 00:33:58.620 --> 00:34:02.800 the fact that the developer has to juggle loops and which loop and that loop. 00:34:02.920 --> 00:34:03.940 No, that's not the right loop. 00:34:04.240 --> 00:34:07.200 Like so many times I've been, Oh, I want to do this request. 00:34:07.640 --> 00:34:10.659 Oh, are you using court or an async web framework? You, 00:34:10.879 --> 00:34:12.840 there is an async loop, but not that one. 00:34:12.899 --> 00:34:14.720 you need the one created by the web framework. 00:34:14.860 --> 00:34:17.460 Like, oh, did you initialize the async database connection 00:34:18.240 --> 00:34:19.700 before it ran the web thing? 00:34:19.720 --> 00:34:21.100 And like, nope, that's the wrong one. 00:34:21.500 --> 00:34:24.320 And there's all this weird juggling that's just so janky. 00:34:25.000 --> 00:34:26.659 And I'm pretty sure it's janky 00:34:27.240 --> 00:34:28.940 because we didn't have true threads. 00:34:29.460 --> 00:34:31.500 And the possibility, oh, what if you crisscross this? 00:34:32.160 --> 00:34:33.340 Everything sort of falls apart. 00:34:33.540 --> 00:34:35.540 And if you could just say, look, it's just multi-threaded. 00:34:35.760 --> 00:34:37.740 And there is a thing that called the loop 00:34:37.780 --> 00:34:38.840 and that's where stuff runs. 00:34:39.300 --> 00:34:41.080 And I don't care who started it. 00:34:41.220 --> 00:34:41.760 It's our loop. 00:34:41.899 --> 00:34:47.020 When I need a loop, the runtime Python itself has a loop for me and I'll use it. 00:34:47.320 --> 00:34:47.419 Right. 00:34:47.720 --> 00:34:49.639 And I feel like you sort of took that philosophy a little bit. 00:34:49.679 --> 00:34:49.800 Right. 00:34:50.320 --> 00:34:50.480 Yeah. 00:34:50.639 --> 00:34:58.180 I'd say like the number one inspiration for Sonio is Tokyo, which is the number one Rust 00:34:58.460 --> 00:34:59.100 asynchronous runtime. 00:35:00.620 --> 00:35:03.600 That's probably because like I spend like in

Rust a bunch of time. 00:35:04.900 --> 00:35:09.900 So yeah, like, I don't know, for example, like the weird thing to me is like, we still like, 00:35:09.980 --> 00:35:14.020 if you just want to do like an asynchronous program in Python, 00:35:14.410 --> 00:35:18.160 like a program, like let's say a script or, I don't know, 00:35:18.320 --> 00:35:21.680 like a simple CLI, we still need to do like a synch.io run. 00:35:22.320 --> 00:35:22.920 Like why? 00:35:23.800 --> 00:35:28.080 Like the, yeah, the amount of things you need to know 00:35:28.680 --> 00:35:33.220 and decide and stuff there is like completely different 00:35:33.440 --> 00:35:34.340 from other languages. 00:35:34.520 --> 00:35:34.600 Yeah. 00:35:34.690 --> 00:35:36.860 Because again, like think about JavaScript. 00:35:37.670 --> 00:35:38.520 You don't decide anything. 00:35:38.580 --> 00:35:40.180 Like that's async in JavaScript. 00:35:40.900 --> 00:35:41.260 That's true. 00:35:41.640 --> 00:35:43.140 Well, I'll tell you a little bit why you can't just, 00:35:43.460 --> 00:35:44.560 why you still got to call async run 00:35:44.720 --> 00:35:46.460 because there's like a foundational layer 00:35:46.980 --> 00:35:47.740 that's not present. 00:35:48.300 --> 00:35:50.120 And it's your job to write the foundation 00:35:50.420 --> 00:35:53.580 on which asyncio executes, right? 00:35:53.960 --> 00:35:55.080 Like it's still your job to figure, 00:35:55.220 --> 00:35:56.820 okay, how do I actually create a loop? 00:35:56.880 --> 00:35:57.640 How do I run a loop? 00:35:57.740 --> 00:35:59.120 How do I, like, it's your job. 00:35:59.520 --> 00:36:01.340 You can't just create an async def method 00:36:01.500 --> 00:36:04.100 and call it because there's nowhere for it to go 00:36:04.280 --> 00:36:05.440 until you go create, you know what I mean? 00:36:05.540 --> 00:36:06.120 There's like, there's just, 00:36:06.800 --> 00:36:08.660 But the foundation isn't quite there. 00:36:08.960 --> 00:36:10.800 But again, it's an entry point, right? 00:36:11.700 --> 00:36:14.080 So in Tonio, for example, yes, you have Tonio run. 00:36:14.210 --> 00:36:18.120 You want to instruct everything and configure the runtime and do whatever you want. 00:36:18.180 --> 00:36:19.340 Yes, you can do that. 00:36:19.550 --> 00:36:20.800 You don't want to do that. 00:36:21.060 --> 00:36:23.600 You have a single decorator, which is Tonio main. 00:36:24.020 --> 00:36:30.780 You put it on your function, your entry point in your program, and it's done. 00:36:32.500 --> 00:36:37.300 Yeah. All right. Before we talk about that, before we get too much in the weeds of it, 00:36:37.400 --> 00:36:43.480 why don't you just give us like, talk us through writing a program in this. There's also two ways, 00:36:43.580 --> 00:36:47.920 like maybe you could, after you talk about this, you could like sort of get into the, 00:36:48.380 --> 00:36:55.600 what this concept of colored functions versus not, and sometimes call async code viral code.

And like, 00:36:55.680 --> 00:37:02.360 let's just like talk through the differences here. Yeah. So, okay. Yes. Tonyo has two 00:37:02.360 --> 00:37:06.940 supports two different syntax modes. 00:37:08.030 --> 00:37:11.960 And that's because, again, when I designed this, 00:37:12.580 --> 00:37:15.180 given that I just throw everything away 00:37:15.670 --> 00:37:16.620 and started from scratch, 00:37:17.290 --> 00:37:18.820 I ended up having a system 00:37:19.150 --> 00:37:23.580 which didn't really require async await syntax 00:37:23.730 --> 00:37:26.440 or a notation because at the end of the day, 00:37:26.550 --> 00:37:27.760 behind async await notation, 00:37:28.060 --> 00:37:29.740 those are generated, right? 00:37:30.740 --> 00:37:45.840 And so given the effort to support like two different syntax was like minimal, writing everything from scratch, I ended up like leaving up to the final user to decide, hey, you dislike async await temptation? 00:37:46.460 --> 00:37:55.400 There's still plenty of people that like, I don't know why, but they argue like all the time about, oh, you know, like I have to put async everywhere and a wait everywhere, whatever. 00:37:56.180 --> 00:37:57.560 OK, you don't like that. 00:37:58.000 --> 00:37:58.320 Guess what? 00:37:58.640 --> 00:38:00.980 Tonya also have just a yield syntax. 00:38:01.200 --> 00:38:10.240 So instead of await, you just yield from the next coroutine you want to suspend for or wait for. 00:38:11.320 --> 00:38:11.800 And that's it. 00:38:11.800 --> 00:38:15.460 You don't need to write async def everywhere. 00:38:17.480 --> 00:38:19.960 To be here is still colored, right? 00:38:20.140 --> 00:38:26.020 Because the moment you have a generator function, then whatever it calls it before has to be a generator. 00:38:26.080 --> 00:38:33.060 as well. But I guess my point was mainly like, okay, to support the syntax doesn't really take 00:38:33.210 --> 00:38:40.160 that much of a work. And once it was settled in, I mean, it's there and people can decide by, 00:38:40.580 --> 00:38:48.120 hey, you dislike one notation? Sure, just use the other one. But yes, let's say for people 00:38:48.580 --> 00:38:55.220 familiar with AsyncIO, the syntax, like the AsyncAwait syntax is very similar to AsyncIO. 00:38:55.900 --> 00:39:06.540 So you basically have your coroutine, so your sync dev, whatever you want, and you await primitives or stuff. 00:39:08.380 --> 00:39:20.800 So the main difference is that instead of importing stuff from a sync.io, so like, I don't know, sleep or timer or whatever, you import similar primitives from Tonio. 00:39:21.440 --> 00:39:28.080 We, of course, have way less primitives because, again, like, as I said before, AsyncIO has too many primitives. 00:39:29.380 --> 00:39:39.660 And the only big difference from AsyncIO, let's say, native people is that you have, like, spawn methods. 00:39:40.260 --> 00:39:41.879 So instead of saying... 00:39:43.460 --> 00:39:47.720 So in AsyncIO, when

you want to make things, 00:39:48.200 --> 00:39:50.200 like spawn several tasks 00:39:50.640 --> 00:39:52.380 and then wait for all of them to complete, 00:39:52.680 --> 00:39:53.500 you usually do... 00:39:53.500 --> 00:39:56.080 You have different ways of doing this, right? 00:39:56.260 --> 00:39:58.620 You can have a task set. 00:39:59.240 --> 00:40:02.780 You can have scopes from Trio. 00:40:04.340 --> 00:40:06.060 But I mean, you can use gather 00:40:06.300 --> 00:40:08.980 so you can create tasks and then gather. 00:40:10.380 --> 00:40:10.840 with SyncIO. 00:40:10.920 --> 00:40:12.800 So Tonio is like, 00:40:12.980 --> 00:40:14.360 to do this is, 00:40:14.820 --> 00:40:16.420 you just have like two methods. 00:40:16.500 --> 00:40:17.160 You either spawn 00:40:19.000 --> 00:40:19.740 asynchronous stuff 00:40:19.980 --> 00:40:21.200 or you spawn something 00:40:21.200 --> 00:40:21.900 that is blocking 00:40:22.460 --> 00:40:23.480 and then the primitive 00:40:24.400 --> 00:40:25.920 is spawn blocking. 00:40:26.980 --> 00:40:28.140 The other major difference 00:40:28.400 --> 00:40:29.000 from a SyncIO 00:40:29.080 --> 00:40:30.680 is that when you spawn something 00:40:31.180 --> 00:40:33.000 that gets run immediately. 00:40:34.560 --> 00:40:35.940 Whereas like in a SyncIO 00:40:35.960 --> 00:40:37.080 this is true only 00:40:37.340 --> 00:40:38.440 if you create a task. 00:40:38.960 --> 00:40:41.380 Otherwise, everything else is eager. 00:40:41.960 --> 00:40:45.900 So in order to run whatever you want to wait, 00:40:45.980 --> 00:40:46.980 you need to await. 00:40:47.300 --> 00:40:49.600 So in AsyncIO, the vast majority of operations 00:40:50.540 --> 00:40:54.100 bound together the launching the operation 00:40:54.380 --> 00:40:55.440 and waiting for the result. 00:40:55.740 --> 00:40:58.320 Unless you create a task that starts immediately 00:40:58.720 --> 00:40:59.880 and then you await later, right? 00:41:00.820 --> 00:41:02.640 In Tonya, everything is wrapped around the spawn 00:41:02.840 --> 00:41:05.440 because you can call spawn and forget about it 00:41:05.660 --> 00:41:08.760 or you can await Tonya spawn to wait for the results, right? 00:41:08.900 --> 00:41:12.000 Or you can park the result of the spawn, the join, and join later. 00:41:12.980 --> 00:41:16.640 So that's, I don't know if, like, maybe it's more confusing than, 00:41:17.380 --> 00:41:19.860 it's hard to explain it simply. 00:41:20.260 --> 00:41:24.680 Maybe, but one of the challenges I've seen with standard asyncIO 00:41:24.780 --> 00:41:28.560 is you want to create a bunch of work and let it run, 00:41:28.820 --> 00:41:30.340 and then you want to get the answers back. 00:41:30.520 --> 00:41:33.140 So a naive way would be, like, call a bunch of stuff, 00:41:34.140 --> 00:41:35.980 and then when you call await, 00:41:36.380 --> 00:41:38.980 It's like you've got to create them not started 00:41:39.280 --> 00:41:40.440 and then start all of them 00:41:40.530 --> 00:41:42.500 and then go through each one and await them. 00:41:42.730 --> 00:41:44.520 Whereas this way, you get a list back 00:41:44.590 --> 00:41:45.680 and you just await them in order 00:41:45.990 --> 00:41:47.640 or await one after another somehow. 00:41:48.020 --> 00:41:49.380 They're already started, right? 00:41:49.470 --> 00:41:51.220 And so there's like the skip of this like, 00:41:51.330 --> 00:41:52.400 okay, I've got all the things 00:41:52.540 --> 00:41:53.540 that are going to become tasks, 00:41:53.660 --> 00:41:54.780 but I got to turn them into tasks 00:41:54.900 --> 00:41:56.180 so then I can await them 00:41:56.320 --> 00:41:57.400 because if I regularly await them, 00:41:57.520 --> 00:41:58.980 it'll still fall back to serial 00:41:59.500 --> 00:42:01.080 just running on the event loop, you know? 00:42:01.340 --> 00:42:01.440 Yeah. 00:42:01.730 --> 00:42:02.920 And I mean, like in Tanya is the same. 00:42:02.990 --> 00:42:04.260 Like if you await a coroutine, 00:42:04.370 --> 00:42:06.340 it means like I want to wait 00:42:06.360 --> 00:42:08.360 that to happen, right? If you don't want to 00:42:08.460 --> 00:42:10.300 just spawn, park 00:42:10.500 --> 00:42:12.220 the result of spawn into a 00:42:12.460 --> 00:42:14.220 variable, await later. I think 00:42:14.340 --> 00:42:15.120 there's an example 00:42:16.420 --> 00:42:17.480 down below in the page 00:42:19.240 --> 00:42:19.720 about 00:42:20.100 --> 00:42:21.960 spawning something and awaiting later. 00:42:24.520 --> 00:42:24.780 Somewhere. 00:42:26.220 --> 00:42:27.420 There's a bunch of documentation. 00:42:28.220 --> 00:42:28.960 Spawning there, yeah. 00:42:30.480 --> 00:42:31.760 But yeah, the idea 00:42:32.200 --> 00:42:34.000 is, and again, I didn't invent 00:42:34.040 --> 00:42:34.820 anything because 00:42:34.840 --> 00:42:41.240 the syntax. Yes, this is the example. So in this example, we compute like numbers, 00:42:41.550 --> 00:42:46.160 like stupid example, but it is to give the point. So we start like two functions, 00:42:46.850 --> 00:42:53.080 which computes numbers. Then we await for third function. So we immediately wait for the third 00:42:53.380 --> 00:43:00.260 result. And then we wait for the first two results. So in the meantime, so when you do the first poem, 00:43:00.660 --> 00:43:05.560 those two functions start already, like immediately at that point in your code, 00:43:06.660 --> 00:43:08.180 which kind of makes sense if you think about it, 00:43:08.230 --> 00:43:12.400 because it's like syntax-based, like you're invoking those coroutines, right? 00:43:12.980 --> 00:43:14.980 In that moment, you just don't await for them. 00:43:15.960 --> 00:43:20.820 And so at the end, like you just await for the parallel, actual parallel result. 00:43:21.220 --> 00:43:25.280 Like everything here is concurrent, but the only point in which you have parallel code, 00:43:25.580 --> 00:43:30.140 the actual parallel code is when you await for parallel, right?

00:43:30.920 --> 00:43:34.160 So that's the major difference from a Synco and Tonio. 00:43:35.000 --> 00:43:39.340 When you spawn stuff, that thing happened in parallel. 00:43:39.700 --> 00:43:43.320 So by default, Tonio starts with the number of threads 00:43:44.140 --> 00:43:46.340 equal to the number of your CPUs. 00:43:47.700 --> 00:43:52.980 The blocking, Tonio also has a blocking pool for blocking stuff. 00:43:54.360 --> 00:43:54.860 But that's separate. 00:43:54.900 --> 00:43:59.200 Let's say the main work loop that runs your code, 00:44:00.460 --> 00:44:04.280 by default, you have X threads per CPU cores. 00:44:05.380 --> 00:44:09.800 So if you run this on your M3 Max, M3 Studio Max, 00:44:10.740 --> 00:44:11.240 what was it? 00:44:11.600 --> 00:44:14.620 You will end up having 18 threads running stuff. 00:44:14.780 --> 00:44:14.920 Yeah. 00:44:15.360 --> 00:44:18.480 Which is kind of what we want in general, right? 00:44:19.300 --> 00:44:20.840 Like if I write a program- 00:44:20.840 --> 00:44:22.140 I'm sure you can configure it, right? 00:44:22.220 --> 00:44:24.140 You could configure the thread pool to say, 00:44:24.220 --> 00:44:26.580 like, look, this thing has to be a good citizen. 00:44:26.820 --> 00:44:28.220 Let's just take the number of CPUs, 00:44:28.320 --> 00:44:29.400 divide by two or something like that 00:44:29.400 --> 00:44:30.460 so I can still do other work. 00:44:30.680 --> 00:44:33.060 Yeah, you can like, again, 00:44:34.080 --> 00:44:36.700 that Tonyo main decorator 00:44:36.980 --> 00:44:38.360 or the Tonyo run method 00:44:40.840 --> 00:44:42.080 accepts some parameters 00:44:42.220 --> 00:44:43.680 so you can configure like the size 00:44:43.700 --> 00:44:45.220 of the standard thread pool, 00:44:45.640 --> 00:44:47.640 the maximum amount of blocking threads 00:44:47.840 --> 00:44:48.680 you want to spawn 00:44:49.280 --> 00:44:51.020 because that does get spawned on demand. 00:44:52.460 --> 00:44:58.660 you can configure if you want to have support for context bars because it's not that's another thing 00:44:58.940 --> 00:45:07.360 about python right like we kind of went from thread locals to context bars uh but in thonio 00:45:07.520 --> 00:45:12.840 that's a bit more complicated right because you have async code in multiple threads yeah because 00:45:12.930 --> 00:45:18.340 they used to leverage the thread right yeah so if you want to actually use context bars 00:45:18.360 --> 00:45:20.900 you have to tell to the runtime, right? 00:45:20.980 --> 00:45:24.360 Because otherwise, like, some side effects might be weird 00:45:24.940 --> 00:45:25.560 in that condition. 00:45:26.620 --> 00:45:26.960 Hold on. 00:45:28.319 --> 00:45:30.240 Nomenclature definition, please, here. 00:45:30.500 --> 00:45:32.160 What are context vars for people who don't know? 00:45:32.260 --> 00:45:32.720 Give us examples. 00:45:32.840 --> 00:45:33.140 Oh, yeah. 00:45:33.540 --> 00:45:37.940 So context vars, so the interpreter has these, 00:45:39.060 --> 00:45:43.280 every thread into the Python interpreter 00:45:43.580 --> 00:45:45.780 has what it's called a context. 00:45:47.140 --> 00:45:51.160 And so with a sync.io, so what was the problem? 00:45:51.500 --> 00:46:01.280 So we used to have thread locals, which meant like if I have two threads and I have, for example, a request, 00:46:01.660 --> 00:46:10.900 and I have to keep the request state global, like global between quotes in my code, 00:46:11.100 --> 00:46:16.900 but I want to use the correct request in the code 00:46:17.810 --> 00:46:21.480 and don't make them mix in the two different threads, states, 00:46:22.460 --> 00:46:26.800 we use thread vocals because that's basically like a dictionary, 00:46:27.210 --> 00:46:30.260 more or less, where you can store that data. 00:46:30.640 --> 00:46:34.200 And that snapshot of data is for a single thread. 00:46:35.060 --> 00:46:39.200 Context var are a similar concept, but for a same code, 00:46:39.340 --> 00:46:42.480 which means you can store global, 00:46:43.440 --> 00:46:44.560 again, between quotes, 00:46:44.900 --> 00:46:50.280 global state from an asynchronous code, 00:46:50.880 --> 00:46:52.880 suspend, and when you get back, 00:46:53.360 --> 00:46:55.780 you have the correct object back 00:46:56.620 --> 00:46:58.600 instead of mixing global state 00:46:58.880 --> 00:46:59.840 between different coroutines. 00:47:00.040 --> 00:47:01.420 Was my explanation good? 00:47:01.760 --> 00:47:02.280 Yeah, that's good. 00:47:02.900 --> 00:47:04.200 People are probably pretty common 00:47:04.380 --> 00:47:05.680 with flask.request. 00:47:05.860 --> 00:47:06.080 Yeah. 00:47:06.120 --> 00:47:07.940 Because you've got a view method 00:47:07.960 --> 00:47:10.500 or maybe deep down inside some lower part of your program, 00:47:10.720 --> 00:47:12.300 you're like, well, I need to know what the URL was. 00:47:12.300 --> 00:47:14.260 You just say Flask.request, you don't pass it around. 00:47:14.460 --> 00:47:17.080 Like very handy, probably architecturally a bad choice. 00:47:17.960 --> 00:47:20.080 You know, it's hard to like test it out, right? 00:47:20.200 --> 00:47:20.700 And you mock it. 00:47:20.720 --> 00:47:21.140 I don't know. 00:47:21.640 --> 00:47:22.660 Emmet does the same. 00:47:23.860 --> 00:47:25.900 It's the one thing I copied from Flask 00:47:26.200 --> 00:47:27.100 because I really like it. 00:47:27.960 --> 00:47:30.880 But I think we can do the same argument about database, right? 00:47:31.060 --> 00:47:32.980 Like with a synchronous database, 00:47:33.140 --> 00:47:35.340 like you have to start a context, 00:47:35.740 --> 00:47:40.860 Like, async with database, whatever, and then you need to pass that, like, all through. 00:47:41.340 --> 00:47:43.440 Like, a context bar is more handy. 00:47:43.840 --> 00:47:44.260 I don't know. 00:47:44.960 --> 00:47:45.340 Yeah, yeah. 00:47:45.380 --> 00:47:47.880 It certainly unlocks some really interesting things. 00:47:48.020 -->

00:47:50.080 Some extensions do cool things with it. 00:47:50.340 --> 00:47:51.300 So that's what you're talking about. 00:47:51.440 --> 00:47:54.960 But because it's not all just tied to the thread and using asyncio context variables, 00:47:55.220 --> 00:47:57.260 now it's kind of shared in potential ways. 00:47:57.540 --> 00:47:59.840 It's a little bit, you got to opt into it. 00:47:59.940 --> 00:48:00.200 That's right? 00:48:00.440 --> 00:48:01.120 Yeah, CSR. 00:48:01.340 --> 00:48:03.560 Again, like, it's a Boolean when you want to start the runtime. 00:48:05.100 --> 00:48:08.740 And I mean, it's an implementation detail. 00:48:09.860 --> 00:48:11.380 So it is just to be like, 00:48:11.700 --> 00:48:13.780 I think the big warning at the beginning 00:48:14.400 --> 00:48:15.040 should have been like, 00:48:15.300 --> 00:48:16.580 Tonyo is very alpha right now. 00:48:18.000 --> 00:48:20.520 So maybe that decision will change in the future 00:48:20.680 --> 00:48:22.820 because I think right now the default is false. 00:48:23.540 --> 00:48:24.940 That might become true 00:48:25.410 --> 00:48:27.580 as soon as I stabilize the API. 00:48:28.420 --> 00:48:29.480 But yeah, the idea was 00:48:29.880 --> 00:48:32.080 if you're testing Tonyo right now, 00:48:32.560 --> 00:48:34.540 if you're using Tonyo to do some tests right now, 00:48:34.880 --> 00:48:41.560 I want the developer to explicitly state, okay, I'm going to use context bars. 00:48:41.960 --> 00:48:48.280 Because again, the fact you have multiple threads and stuff happening in parallel, 00:48:49.200 --> 00:48:55.300 I just want for the developer to be sure if it's thinking model is correct 00:48:55.460 --> 00:48:57.820 before trying to do some stuff. 00:48:59.160 --> 00:49:01.060 Because again, Tony is multi-threaded. 00:49:01.160 --> 00:49:07.900 multi-threaders is usually like not it's it's i don't know i think for humans multi-threading is 00:49:08.200 --> 00:49:16.520 hard to get sometimes um so that's the trade-off right like uh yes we now have all of these 00:49:16.700 --> 00:49:22.480 capabilities on the other end we need to think about the fact that the sync io always had this 00:49:22.940 --> 00:49:30.420 uh hidden uh let's say a feature which is hey there's only one coroutine running at the time 00:49:30.620 --> 00:49:32.640 Whereas Antonio can have like whatever number. 00:49:32.940 --> 00:49:33.100 Yep. 00:49:34.160 --> 00:49:37.160 Now you still had to manage stuff across await calls. 00:49:37.760 --> 00:49:43.520 So that was, I feel like people felt like the GIL saved them from all thread considerations, 00:49:43.960 --> 00:49:46.200 like locks and semaphores and so on. 00:49:46.460 --> 00:49:47.360 And I don't think it did. 00:49:47.640 --> 00:49:50.780 There's not a guarantee that like a thread couldn't interrupt you. 00:49:50.880 --> 00:49:52.780 It's just less likely to interrupt you, right? 00:49:53.080 --> 00:49:59.600 So like the simple thing is that it did around other primitives. 00:49:59.940 --> 00:50:08.680 So for example, in a SyncIO, it's really hard to deadlock yourself because you're using like a threading lock. 00:50:09.000 --> 00:50:16.000 Because again, like the hidden feature of a SyncIO is you have only one thread working on that shit, right? 00:50:16.740 --> 00:50:17.320 Yeah, exactly. 00:50:17.740 --> 00:50:19.080 Like you can do all the locks you want. 00:50:19.160 --> 00:50:20.600 It's the same thread in the re-entrance. 00:50:21.140 --> 00:50:21.380 Exactly. 00:50:21.920 --> 00:50:24.500 It's effectively a no-op other than it costs CPU. 00:50:24.920 --> 00:50:25.640 Yes, exactly. 00:50:25.860 --> 00:50:38.300 Whereas in Tonio, like that's exactly where people usually make mistakes because now you have the Tonio locking because you have the asynchronous locks, but you also have thread locks, which are two different things. 00:50:39.000 --> 00:50:42.800 And if you, and so like you can deadlock Tonio. 00:50:43.100 --> 00:50:43.300 How? 00:50:43.580 --> 00:50:43.700 Yeah. 00:50:44.020 --> 00:50:44.940 Welcome to multi-threading. 00:50:45.280 --> 00:50:45.480 Yes. 00:50:45.660 --> 00:50:52.160 Like you, you, you, you create, like you enter a threading lock and then you await a coroutine inside the threading lock. 00:50:52.310 --> 00:50:53.320 Now you're deadlocked. 00:50:53.580 --> 00:50:53.680 Yeah. 00:50:53.780 --> 00:50:55.860 Because the two things doesn't speak each other. 00:50:56.960 --> 00:50:59.700 But I guess my advice, 00:51:00.060 --> 00:51:03.360 if people want to start working in multi-threaded Python, 00:51:03.540 --> 00:51:08.620 my advice is just think about those two rules 00:51:09.140 --> 00:51:10.860 and then multi-threading is not that hard. 00:51:11.080 --> 00:51:12.380 Again, rule number one, 00:51:13.060 --> 00:51:16.800 never await inside a threading lock. 00:51:17.180 --> 00:51:17.960 That's a simple rule. 00:51:18.660 --> 00:51:21.740 Also because in Python, you use width lock. 00:51:22.240 --> 00:51:25.400 So that's super easy even to identify, right? 00:51:26.340 --> 00:51:27.620 That's rule number one. 00:51:28.120 --> 00:51:30.160 Rule number two, you have multiple threads. 00:51:30.510 --> 00:51:32.700 So if you want to write into anything, 00:51:33.760 --> 00:51:37.220 I mean, all the objects are thread safe in free thread of Python 00:51:37.420 --> 00:51:40.560 because a dictionary is thread safe, a list is thread safe, whatever. 00:51:41.080 --> 00:51:45.840 But that doesn't mean like you can screw up in using those. 00:51:46.020 --> 00:51:48.780 Because if you write state anywhere, 00:51:49.400 --> 00:51:58.560 You might have like race conditions and side effects because you expect to write something and then expect to read that value. 00:51:58.800 --> 00:52:01.380 But in the meantime, you have another thread right into the same location. 00:52:01.680 --> 00:52:08.180 So again, like my advice is rule number one, never use locks, threading locks, and then await.

00:52:08.580 --> 00:52:12.560 If you want to have an async lock, then use an async lock. 00:52:12.560 --> 00:52:13.880 So a Tonio lock for that. 00:52:14.720 --> 00:52:16.620 Rule number two, watch out. 00:52:16.820 --> 00:52:25.340 So you probably need locks when you want to be sure that, like, when you have a sequence of operation, right, on shared memory. 00:52:25.740 --> 00:52:26.880 Those are the two rules. 00:52:27.720 --> 00:52:31.740 I think with those two rules, multithreading is not that hard. 00:52:31.960 --> 00:52:32.440 Yeah, agreed. 00:52:32.660 --> 00:52:38.560 So does Tonio come with its own dedicated lock and synchronization primitives? 00:52:38.900 --> 00:52:41.460 Yes, there's an entire sync module. 00:52:41.980 --> 00:52:45.600 So Tonio has different modules. 00:52:46.580 --> 00:52:51.920 Tonya time, which contains time primitives, like timeouts, timers, etc. 00:52:52.300 --> 00:52:56.400 It has async module, which contains all the synchronization primitives, 00:52:56.580 --> 00:52:59.520 so logs, semaphore, barriers, etc. 00:53:01.000 --> 00:53:06.380 It has the network module, which is one providing like sockets. 00:53:06.960 --> 00:53:10.580 So that's quite a big difference from a sync.io. 00:53:10.580 --> 00:53:16.080 So in a sync.io, we don't have like an async socket library. 00:53:16.440 --> 00:53:20.620 So if you think about the standard lib socket module, 00:53:21.359 --> 00:53:25.480 Tonio, AsyncIO needs you to use that 00:53:25.760 --> 00:53:28.280 and then pass the sockets to the event loop 00:53:28.560 --> 00:53:30.320 and then handle the socket through the event. 00:53:31.200 --> 00:53:34.360 Whereas in Tonio, there's a network module 00:53:34.540 --> 00:53:36.340 and inside that module, there's a socket module, 00:53:36.670 --> 00:53:40.380 which has the same exact interface of the standard lib plus socket. 00:53:40.960 --> 00:53:42.040 The only difference is, of course, 00:53:42.600 --> 00:53:45.500 every async method is async in Tonio. 00:53:47.280 --> 00:53:47.640 there's the 00:53:48.860 --> 00:53:50.140 streams module 00:53:50.240 --> 00:53:52.140 inside the network module which is the high level 00:53:53.120 --> 00:53:53.900 trio like 00:53:54.580 --> 00:53:55.740 interface to 00:53:56.160 --> 00:53:57.440 make it easy to manage 00:53:57.980 --> 00:54:00.020 network. It has a file 00:54:00.140 --> 00:54:02.020 system module which 00:54:02.280 --> 00:54:02.640 contains 00:54:03.980 --> 00:54:04.640 all the standard 00:54:06.440 --> 00:54:07.880 open kind of 00:54:08.220 --> 00:54:09.060 methods that they 00:54:09.240 --> 00:54:10.300 async with open 00:54:10.860 --> 00:54:12.940 tonio.files.open something like that 00:54:13.100 --> 00:54:14.539 yeah.fs yes 00:54:15.260 --> 00:54:16.500 and it also contains like 00:54:17.400 --> 00:54:18.720 mirror of the 00:54:19.080 --> 00:54:20.260 pathlib, standard lib 00:54:20.680 --> 00:54:22.100 pathlib, so you can use like 00:54:22.400 --> 00:54:24.780 frontoniofs import path and that path 00:54:24.940 --> 00:54:26.060 has all the methods 00:54:26.700 --> 00:54:28.280 that are required to be asynchronous 00:54:29.840 --> 00:54:30.860 so that's 00:54:31.020 --> 00:54:32.120 more or less like the 00:54:33.000 --> 00:54:33.640 design 00:54:34.860 --> 00:54:36.260 I don't think like 00:54:36.420 --> 00:54:38.200 so right now I'm at 00:54:38.360 --> 00:54:40.680 tonio 0.9.14 00:54:40.960 --> 00:54:41.440 I guess 00:54:42.580 --> 00:54:44.520 and I don't think 00:54:44.540 --> 00:54:48.660 There's much left in terms of modules and primitives and features. 00:54:50.440 --> 00:54:52.280 The big feature missing is Windows support. 00:54:52.740 --> 00:54:53.440 More on that later. 00:54:55.839 --> 00:55:02.020 But yeah, I'd say I'm close to stabilize the API. 00:55:02.380 --> 00:55:04.440 So right now, again, this is all alpha. 00:55:05.040 --> 00:55:07.780 That's mostly because there's only me working on this. 00:55:08.920 --> 00:55:12.300 And I have Fable working on a bunch of other things, 00:55:12.920 --> 00:55:16.480 which I guess like if I can digress for a second, 00:55:17.040 --> 00:55:20.420 I think like I have a very peculiar and unique way of using AI 00:55:20.840 --> 00:55:22.020 because right now I'm like, 00:55:22.780 --> 00:55:26.080 and I guess like it's a good use case to advertise. 00:55:26.420 --> 00:55:28.180 So the entirety of Tonio, 00:55:28.340 --> 00:55:32.160 so Tonio has maybe 1% AI written code. 00:55:32.380 --> 00:55:35.900 The only code that was written by Claude 00:55:36.780 --> 00:55:38.140 was the pytest plugin. 00:55:38.540 --> 00:55:40.520 Oh, by the way, we also have a pytest plugin. 00:55:40.780 --> 00:55:42.760 so you can hide the smart tone. 00:55:45.200 --> 00:55:48.220 But that's the only part that I let AI touch. 00:55:49.640 --> 00:55:51.800 Everything else, like it's old school. 00:55:52.210 --> 00:55:56.780 I written all of it like the old way, text editor, 00:55:57.440 --> 00:56:01.280 not even any suggestion and everything. 00:56:02.420 --> 00:56:07.260 What I used AI for, and it helped me a lot 00:56:07.520 --> 00:56:12.660 into reaching a state in which I think Tonio is really stable now 00:56:13.520 --> 00:56:18.480 was to use AI to build projects that were using Tonio 00:56:18.860 --> 00:56:21.440 and stress test everything about Tonio. 00:56:21.510 --> 00:56:24.400 So for example, I made a port of Pi, 00:56:25.570 --> 00:56:27.520 the harness from Maria Zeckner. 00:56:28.510 --> 00:56:31.440 So I made Claude rewrite the whole thing in Python 00:56:32.200 --> 00:56:35.900 because Pi is written in TypeScript, regardless of the name. 00:56:37.320 --> 00:56:40.960 So I made that entire harness with Claude saying, 00:56:41.030 --> 00:56:42.020 okay, take five, 00:56:42.660 --> 00:56:46.380 meter every behavior and just use Tonio to do everything. 00:56:47.430 --> 00:56:49.200 And that like helped me like catch 00:56:49.510 --> 00:56:52.740 like probably 95% of the bugs

in Tonio. 00:56:53.720 --> 00:56:55.920 And I kind of started doing the same for the ecosystem. 00:56:57.100 --> 00:56:59.320 Maybe we can talk about it a bit later, 00:56:59.740 --> 00:57:01.760 but yeah, that's also like a super, 00:57:02.480 --> 00:57:03.940 it's super useful for me, right? 00:57:04.080 --> 00:57:06.240 Like as a solo open source developer, 00:57:06.620 --> 00:57:08.660 which, yeah, I want to focus on Tonio. 00:57:09.080 --> 00:57:11.500 I don't have time to do all the other packages. 00:57:12.040 --> 00:57:13.320 And so, yeah, that was nice. 00:57:13.440 --> 00:57:14.760 That's a very interesting way. 00:57:14.760 --> 00:57:16.080 I hadn't really considered that. 00:57:16.300 --> 00:57:19.360 It's just like, I need users before I have users. 00:57:19.840 --> 00:57:21.840 I need these three use cases covered. 00:57:21.980 --> 00:57:24.240 So give me a web app that uses an async database. 00:57:24.500 --> 00:57:27.180 Give me this terminal app and so on. 00:57:27.180 --> 00:57:28.360 And that's, yeah, that's really cool. 00:57:28.720 --> 00:57:28.920 Yeah. 00:57:29.660 --> 00:57:29.800 Okay. 00:57:30.180 --> 00:57:30.620 I agree. 00:57:30.820 --> 00:57:31.700 Ecosystem is interesting. 00:57:31.780 --> 00:57:32.660 I want to talk about it. 00:57:32.840 --> 00:57:34.940 But before that, I want to come back to blocking threads. 00:57:35.420 --> 00:57:35.680 Yes. 00:57:35.960 --> 00:57:39.600 So blogging threads are CPU bound generally type of things. 00:57:39.610 --> 00:57:40.060 Is that right? 00:57:40.580 --> 00:57:44.680 So I think Tonio is a bit peculiar in this regard 00:57:45.120 --> 00:57:47.680 because it really depends what your program does. 00:57:48.160 --> 00:57:54.020 Because if your program, whatever that is, has mixed load, 00:57:54.210 --> 00:57:58.180 which means like you have both IO bound stuff, 00:57:58.700 --> 00:58:03.960 like network, disk, terminal, whatever, and CPU bound stuff, 00:58:05.180 --> 00:58:10.460 then the only way to be sure that everything keeps running smoothly, 00:58:10.820 --> 00:58:13.300 like on the network part, the IO part, 00:58:13.620 --> 00:58:18.840 is to spawn the CPU bound stuff on the blocking thread pool. 00:58:19.460 --> 00:58:24.740 But if your program is just CPU bound, then who cares? 00:58:24.990 --> 00:58:28.580 Just use the normal thread pool 00:58:29.710 --> 00:58:31.700 because then you don't have anything to block. 00:58:32.000 --> 00:58:34.360 like your CPU bound, your CPU limited. 00:58:35.280 --> 00:58:37.340 So the standard configuration of Tonio 00:58:37.800 --> 00:58:40.760 gives you exactly the perfect number of CPU cores to use. 00:58:41.860 --> 00:58:45.520 On the other hand, if you just have like IO bound. 00:58:46.100 --> 00:58:48.920 So again, like if you have just CPU bound work 00:58:49.100 --> 00:58:50.440 or just IO bound work, 00:58:50.500 --> 00:58:52.540 you can use like the standard pool, 00:58:53.860 --> 00:58:55.200 nothing particular like this. 00:58:55.440 --> 00:58:57.140 The blocking thread pool is useful 00:58:57.440 --> 00:58:59.200 only when you have mixed workloads 00:58:59.820 --> 00:59:01.700 and you need to balance them, right? 00:59:01.960 --> 00:59:03.820 Okay, how do I create such a thing? 00:59:04.160 --> 00:59:07.440 How do I start a task in one or the other? 00:59:07.980 --> 00:59:11.060 So every await or yell destruction 00:59:11.700 --> 00:59:12.780 or a tonio spawn destruction 00:59:13.480 --> 00:59:15.600 runs on the standard thread pool. 00:59:16.660 --> 00:59:19.280 To spawn something on the blocking thread pool 00:59:19.440 --> 00:59:20.860 is just tonio spawn blocking. 00:59:21.190 --> 00:59:21.300 Done. 00:59:21.670 --> 00:59:23.640 You don't need to manage the size of the pool. 00:59:23.690 --> 00:59:26.140 You don't need to create the pool. 00:59:27.480 --> 00:59:32.700 If you want to control concurrency in the blocking threat pool, 00:59:34.240 --> 00:59:37.200 you either use, like, barriers when you spawn stuff 00:59:38.320 --> 00:59:40.260 or semaphores when you spawn stuff. 00:59:41.700 --> 00:59:42.820 But, yeah, like, in general, 00:59:43.300 --> 00:59:45.740 it's not different from spawning anything else. 00:59:46.160 --> 00:59:46.340 Sure. 00:59:46.620 --> 00:59:47.640 So you can spawn blocking. 00:59:48.540 --> 00:59:49.640 Here's what I was thinking. 00:59:49.800 --> 00:59:50.620 Here's what I was getting at. 00:59:51.440 --> 00:59:53.540 That I think would be a sweet feature. 00:59:53.980 --> 00:59:57.100 You've got at tonio.main for the entry point, right? 00:59:57.560 --> 00:59:57.700 Yes. 00:59:58.260 --> 01:00:01.640 So that runs on tonio.run instead of just Python run or whatever. 01:00:02.000 --> 01:00:05.200 For functions that you know are computational, 01:00:05.440 --> 01:00:08.140 you would like to ensure that they run on a blocking thread 01:00:08.640 --> 01:00:09.820 or along those lines. 01:00:10.120 --> 01:00:15.540 You could put a decorator that just says tonio.cpu or whatever. 01:00:15.640 --> 01:00:16.000 I don't know. 01:00:16.120 --> 01:00:19.000 You come up with a name, but that way I don't even have to think. 01:00:19.400 --> 01:00:21.840 If I import another library also built on tonio, 01:00:22.040 --> 01:00:25.020 I don't know, oh, that function wants me to run it this way. 01:00:25.490 --> 01:00:26.480 No, I just call it. 01:00:26.640 --> 01:00:29.600 Antonio goes, ah, this is an async CPU one. 01:00:29.910 --> 01:00:33.680 So we'll scale it up, scaffold it up on that, basically. 01:00:33.960 --> 01:00:36.200 That's a nice suggestion, okay? 01:00:36.940 --> 01:00:37.200 All right. 01:00:37.320 --> 01:00:40.200 I think it will land in Antonio 0.10, yes. 01:00:40.580 --> 01:00:41.300 Awesome, okay, cool. 01:00:41.680 --> 01:00:42.380 I like that. 01:00:43.260 --> 01:00:46.720 When you're in the function, the locality of the information 01:00:46.790 --> 01:00:49.080 that, oh, this is computational and so on is right there.

01:00:49.170 --> 01:00:50.920 But when you're calling, and especially in a big app, 01:00:51.090 --> 01:00:51.700 you don't know, right? 01:00:51.780 --> 01:00:52.680 So I think that would be really cool. 01:00:53.060 --> 01:00:53.880 So I'm glad you like it. 01:00:54.060 --> 01:00:55.080 No, we'll work on that. 01:00:55.640 --> 01:00:55.940 Awesome. 01:00:56.940 --> 01:00:58.620 All right, we got time for two really quick things. 01:00:59.040 --> 01:01:02.100 So one thing that is nice about standard asyncio 01:01:02.280 --> 01:01:03.640 is everyone uses asyncio 01:01:03.840 --> 01:01:05.460 if they're writing async code at the moment, 01:01:05.930 --> 01:01:06.840 unless they do something like Trio 01:01:06.970 --> 01:01:07.680 or one of these other things. 01:01:08.060 --> 01:01:11.760 So for example, if I use HTTPX2 or HTTPX 01:01:12.300 --> 01:01:14.520 and I, with async, you know, 01:01:14.840 --> 01:01:16.440 async with create a client of it 01:01:16.500 --> 01:01:17.520 and I await its get, 01:01:18.260 --> 01:01:20.180 that runs on the same sort of deal. 01:01:20.440 --> 01:01:23.040 but it's probably not compatible with Tonio, is it? 01:01:23.270 --> 01:01:24.340 And if it's not, what do I do? 01:01:24.940 --> 01:01:25.760 Yeah, it's not. 01:01:28.580 --> 01:01:32.080 So I have one thing that is the current state, 01:01:32.380 --> 01:01:35.580 and I also have the idea for the future. 01:01:35.850 --> 01:01:40.060 So right now we have a package which is called TonioMonkey, 01:01:40.460 --> 01:01:43.140 and you probably can guess what it means, what it does. 01:01:43.820 --> 01:01:48.180 It basically like monkey patches popular, 01:01:48.860 --> 01:01:55.520 Or I would either say, like, asyncio native libraries I thought about in the last few months. 01:01:56.580 --> 01:01:57.480 That might be useful. 01:01:59.320 --> 01:02:04.980 And so I guess today we have, like, TonyaMonkey can monkey patch async. 01:02:05.320 --> 01:02:05.860 No, sorry. 01:02:06.640 --> 01:02:07.200 PsychoPG. 01:02:08.380 --> 01:02:11.200 It can patch HTTPX and HTTPX2. 01:02:12.200 --> 01:02:13.760 It can patch Redis. 01:02:15.220 --> 01:02:17.240 And I think that's most of it for the moment. 01:02:17.760 --> 01:02:21.040 But I'm open for issues. 01:02:21.360 --> 01:02:28.440 If anyone from the listeners want to try Tonio to an existing code base 01:02:28.680 --> 01:02:32.640 and you have a package you would like to see monkey patched, 01:02:33.180 --> 01:02:35.340 just open an issue in the TonioMonkey repo 01:02:35.900 --> 01:02:39.900 or even in Tonio discussions, and I will look at it. 01:02:41.560 --> 01:02:46.860 Because again, right now I need the vast majority of possible use cases 01:02:46.880 --> 01:02:48.480 to see that I covered everything. 01:02:50.140 --> 01:02:52.460 Ideally, as soon as I stabilize the API 01:02:52.680 --> 01:02:55.040 and I say, okay, this is the thing, 01:02:56.200 --> 01:03:03.380 my idea was to try to contact the NEIO maintainers 01:03:04.900 --> 01:03:13.260 to see if we can have NEIO backend in NEIO, 01:03:13.680 --> 01:03:19.820 which should be like a way easier way to deal with this. 01:03:20.900 --> 01:03:22.720 I'm not sure how much it's feasible because again, 01:03:22.900 --> 01:03:27.540 like the main problem for Tonio right now is that it doesn't support any of the 01:03:27.740 --> 01:03:28.600 synchio primitives. 01:03:28.880 --> 01:03:33.260 So if any code base like relies a ton on task or futures or whatever, 01:03:34.660 --> 01:03:36.520 that's not in the runtime. 01:03:36.940 --> 01:03:37.060 Yeah. 01:03:37.300 --> 01:03:37.420 Yeah. 01:03:37.760 --> 01:03:37.880 Okay. 01:03:38.400 --> 01:03:38.640 Interesting. 01:03:38.980 --> 01:03:39.140 Interesting. 01:03:39.340 --> 01:03:40.560 But there are some choices. 01:03:40.820 --> 01:03:42.580 And are you also creating your own? 01:03:42.740 --> 01:03:48.320 I saw some custom HTTP or some other library or two. 01:03:48.980 --> 01:03:53.640 I'm also trying to be a very contained ecosystem of libraries, 01:03:54.120 --> 01:03:55.460 mainly on the web part. 01:03:55.660 --> 01:04:01.580 So right now, I published a low-level HTTP library for Python 01:04:02.430 --> 01:04:06.440 that is compatible both with Asyncio and Tonio. 01:04:07.040 --> 01:04:08.680 It's called HTTPunk. 01:04:09.100 --> 01:04:17.740 I published an eye-level client, HTTP client, on top of that, which is called Punk Rec. 01:04:19.360 --> 01:04:27.020 So again, you can use that regardless of Tonio Asincaio, because the adaptation layer is inside the package. 01:04:27.580 --> 01:04:38.340 I kind of published a very beta experiment of a Uvicorn port to work in Tonio, 01:04:38.680 --> 01:04:45.120 Which I guess brings up a question like, hey, when will Granian support this? 01:04:45.430 --> 01:04:45.960 Am I right? 01:04:46.220 --> 01:04:46.560 Yeah, right. 01:04:47.640 --> 01:04:48.720 So I have plans. 01:04:50.080 --> 01:04:52.700 It's actually way bigger plans for Granian. 01:04:53.820 --> 01:04:58.700 So as I said before, 2026 was a bit boring in Granian. 01:04:58.900 --> 01:05:02.340 Like, no, a bunch of new features, not a bunch of that stuff. 01:05:02.390 --> 01:05:08.140 But that's just because several months ago, I started designing two things for Granian. 01:05:08.280 --> 01:05:13.140 So the first is this revision two of the ArchGee protocol, 01:05:13.680 --> 01:05:17.020 which will be async independent. 01:05:17.600 --> 01:05:20.580 So it will become like a callback protocol 01:05:21.780 --> 01:05:25.720 because the other big feature I want to add in Granian 01:05:26.180 --> 01:05:29.320 is to support a bunch of runtimes. 01:05:29.660 --> 01:05:31.620 So my idea is not the Granian tree, 01:05:32.240 --> 01:05:36.040 when it will land, no promises on the time here, 01:05:36.380 --> 01:05:37.640 but like when it will land, 01:05:37.980 --> 01:05:45.700 Ideally, it will support Asyncio, Trio, Tonio, gEvent, eventlet, and a bunch of stuff.

01:05:45.890 --> 01:05:55.360 So that's the idea, is to isolate, let's say, the protocol implementation and the Python runtime implementation. 01:05:55.620 --> 01:05:57.540 So it will happen one day. 01:05:59.660 --> 01:06:04.620 In the meantime, I wanted to have a test server to use with Tonio and see what happens. 01:06:04.890 --> 01:06:05.640 So that's why. 01:06:07.220 --> 01:06:08.420 Yeah, using something like 01:06:08.780 --> 01:06:10.860 Cord or FastAPI or something like that. 01:06:11.120 --> 01:06:12.900 If you were able to run it in Granian 01:06:13.400 --> 01:06:15.020 with the Tonio, Tonio 01:06:15.520 --> 01:06:17.220 backend, the async 01:06:17.560 --> 01:06:18.700 endpoints you write would be 01:06:19.120 --> 01:06:19.960 running on Tonio, yeah? 01:06:20.240 --> 01:06:21.360 Yeah, that's the plan. 01:06:22.800 --> 01:06:25.340 In TonioMonkey, there are also patches for FastAPI, 01:06:25.480 --> 01:06:26.820 by the way. And I'm working 01:06:27.060 --> 01:06:28.480 on a one on Django. 01:06:28.920 --> 01:06:30.760 I need to ping 01:06:30.900 --> 01:06:31.900 Carleton on that 01:06:32.820 --> 01:06:34.380 because we met this year 01:06:34.420 --> 01:06:36.400 at this year, PyCon Italy, 01:06:36.940 --> 01:06:38.040 and we talk about it. 01:06:39.259 --> 01:06:42.240 So maybe in a month or so, 01:06:42.660 --> 01:06:44.440 we will also have like a patch for Django. 01:06:44.720 --> 01:06:44.880 Sweet. 01:06:45.190 --> 01:06:51.360 Yeah, I just had Carlton on to talk about 01:06:52.260 --> 01:06:55.700 basically all the async work in Django 6 and 6.0 and so on. 01:06:55.730 --> 01:06:56.360 So that was fun. 01:06:56.400 --> 01:06:58.600 I watched the episode and I was like, 01:06:58.900 --> 01:07:01.420 huh, sounds familiar, huh, huh. 01:07:01.960 --> 01:07:02.980 Yeah, we've talked about that. 01:07:03.030 --> 01:07:03.380 I see. 01:07:03.820 --> 01:07:04.060 Interesting. 01:07:04.480 --> 01:07:04.660 All right. 01:07:04.720 --> 01:07:07.520 Well, I got one final main topic for us. 01:07:07.860 --> 01:07:08.020 Yep. 01:07:08.280 --> 01:07:08.680 Oh, yeah. 01:07:08.720 --> 01:07:10.200 What about windows? 01:07:12.260 --> 01:07:12.520 So. 01:07:12.800 --> 01:07:15.180 And I'm not putting pressure on you to do it. 01:07:15.260 --> 01:07:15.700 No, no, no. 01:07:16.060 --> 01:07:17.560 I'm just rounding out the conversation. 01:07:18.040 --> 01:07:18.300 Okay. 01:07:18.660 --> 01:07:26.480 So if you ask me like a few months ago, my answer would probably be like that short video 01:07:26.920 --> 01:07:31.740 extract from a series like where you had the actor. 01:07:32.780 --> 01:07:32.940 Whatever. 01:07:33.080 --> 01:07:41.700 um you know i'm gonna not gonna not gonna swear on the podcast but uh uh you get what i mean 01:07:41.800 --> 01:07:48.640 uh so i i'm not so first of all like the usual problem so supporting windows in granian 01:07:49.180 --> 01:07:55.220 it's a pain man like a real pain like i'm not joking right because it has

like all these 01:07:55.560 --> 01:08:02.500 quirk behaviors not just in python not just in terms of threads like do you know that win that 01:08:02.520 --> 01:08:11.220 that windows have as like a bug from i think it was like nt4 uh that like if you if you share a 01:08:11.430 --> 01:08:17.859 socket it becomes blocking and so you cannot use it like anywhere um so windows for start like 01:08:18.029 --> 01:08:26.299 windows is a pain and is a very weird operating system second i don't have like a windows environment 01:08:27.440 --> 01:08:29.120 because finally in 2026, 01:08:29.380 --> 01:08:32.200 I was able to finally have just macOS and Linux 01:08:32.580 --> 01:08:34.980 on all my machines, also for gaming. 01:08:35.319 --> 01:08:36.580 So I'm so happy. 01:08:37.960 --> 01:08:39.420 So those are, like I say, 01:08:39.759 --> 01:08:43.240 the two main preambles on Windows support. 01:08:44.499 --> 01:08:48.359 But also because implementing the library 01:08:48.400 --> 01:08:52.120 I use Antonio to manage the Polar, 01:08:52.339 --> 01:08:53.819 the actual event loop, 01:08:54.859 --> 01:08:58.220 It supports Windows, but you know, weird way. 01:08:58.580 --> 01:09:03.520 So, months ago I was like almost a hundred percent convinced now, 01:09:03.839 --> 01:09:08.799 never windows, never now, thanks to fable by chatting with fable. 01:09:09.240 --> 01:09:18.380 I think we found a way to hack into, into meal, this library to, to fake 01:09:18.779 --> 01:09:20.140 some stuff about windows. 01:09:20.220 --> 01:09:24.359 So it will interpret some stuff as TCP sockets. 01:09:24.460 --> 01:09:27.540 even if they are like file descriptors or anything. 01:09:28.140 --> 01:09:30.980 So I won't make a promise here 01:09:31.380 --> 01:09:33.900 because then I have to do it. 01:09:34.240 --> 01:09:37.759 But ideally, end of the year, 01:09:38.120 --> 01:09:41.020 we might see partial Windows support in Tom. 01:09:41.190 --> 01:09:41.319 Sweet. 01:09:41.640 --> 01:09:43.420 So that's the idea right now. 01:09:44.120 --> 01:09:47.640 I still hate Microsoft, to be clear here. 01:09:47.640 --> 01:09:49.740 Like, man, I just like Windows. 01:09:51.720 --> 01:09:54.340 So if any of the listeners, 01:09:54.450 --> 01:09:56.860 like if you're programming in Windows 01:09:57.560 --> 01:09:58.740 and not using Weasel, 01:09:59.160 --> 01:10:00.720 please put down in the comments why. 01:10:01.220 --> 01:10:04.960 Like, why are you self-inflicting this to you? 01:10:05.140 --> 01:10:07.440 Like, explain me the rationale 01:10:08.100 --> 01:10:11.180 into dealing with all the BS 01:10:11.480 --> 01:10:13.200 that Windows 11 is nowadays. 01:10:14.620 --> 01:10:15.580 So yeah, but anyways, 01:10:17.080 --> 01:10:18.440 I will try end of the year 01:10:18.440 --> 01:10:21.120 to have some form of Windows support in time. 01:10:21.700 --> 01:10:23.480 I can tell, I can sense your excitement. 01:10:23.880 --> 01:10:25.800 My Windows computer over there is Windows 10, by the way. 01:10:26.220 --> 01:10:27.520 But I do think for people listening, 01:10:28.240 --> 01:10:30.600 Windows Subsystem for Linux is an escape hatch 01:10:30.950 --> 01:10:31.860 that you can do, right? 01:10:32.020 --> 01:10:33.920 If you had a project, you really want to use this here, 01:10:34.100 --> 01:10:35.840 like set up Windows Subsystem for Linux 01:10:35.950 --> 01:10:36.620 and run it there, right? 01:10:36.840 --> 01:10:40.320 To be here, like the only reason I want to, 01:10:40.500 --> 01:10:42.680 like I want to try to add Windows support in Tonio 01:10:42.820 --> 01:10:43.920 is not for developers, 01:10:44.420 --> 01:10:47.880 is actually to provide developers a way to say, 01:10:48.050 --> 01:10:50.300 okay, I want to build a program in Python 01:10:50.300 --> 01:10:53.180 with Tonio that has to work like everywhere. 01:10:53.700 --> 01:10:54.240 Because right now, again, 01:10:54.440 --> 01:10:56.280 like if you code anything in Tonio, 01:10:56.700 --> 01:10:58.480 it works only on POSIX systems, 01:10:58.640 --> 01:10:59.860 so only Linux and macOS. 01:11:00.280 --> 01:11:02.100 So that's the reason why I want to have 01:11:02.300 --> 01:11:04.360 at least some form of Windows support 01:11:04.480 --> 01:11:06.620 because, I mean, it's not about developers. 01:11:07.340 --> 01:11:09.080 It's about like where the software runs 01:11:09.400 --> 01:11:09.940 at the end of the day. 01:11:10.100 --> 01:11:10.520 Right, right. 01:11:10.620 --> 01:11:12.180 It's the deployment targets, yeah. 01:11:12.560 --> 01:11:12.640 Yeah. 01:11:13.040 --> 01:11:14.360 All right, Joe, final call to action. 01:11:14.620 --> 01:11:15.460 People have been listening 01:11:15.560 --> 01:11:17.000 and they're like, this sounds pretty excellent. 01:11:17.600 --> 01:11:18.380 I want to try it. 01:11:18.740 --> 01:11:19.300 What do you tell them? 01:11:19.540 --> 01:11:27.760 I mean, like, again, I think we're living in the perfect moment to try stuff. 01:11:28.780 --> 01:11:33.900 We've given with this big opportunity, big new opportunity of free-threaded Python. 01:11:35.140 --> 01:11:36.960 We have AI agents. 01:11:38.500 --> 01:11:44.040 And by the way, I was surprised, but like pointing any agent, any model, 01:11:44.780 --> 01:11:48.620 frontier model to a project and say, okay, implement this in Tonio. 01:11:48.840 --> 01:11:49.480 I was surprised. 01:11:49.620 --> 01:11:52.480 They got all the API correctly most of the times, 01:11:53.660 --> 01:11:54.220 which is amazing. 01:11:54.480 --> 01:11:56.580 There's no training data about Tonio. 01:11:56.940 --> 01:12:01.360 And regardless, it was able to drive through stuff. 01:12:02.440 --> 01:12:04.880 So I guess it's the perfect moment to try stuff. 01:12:06.820 --> 01:12:11.440 I'd say if for any reason you're not happy with Asyncio, 01:12:11.700 --> 01:12:18.420 if for any reason you experience any of the pain points we talked about, 01:12:18.740 --> 01:12:23.460 I think it's like the perfect

moment to try new stuff. 01:12:24.040 --> 01:12:28.440 And if you try Tonio and you have like a use case 01:12:29.000 --> 01:12:33.040 where it's not working or the API can be improved, 01:12:33.280 --> 01:12:35.540 like open an issue, open a discussion, 01:12:36.260 --> 01:12:39.120 ping me, DM me on Twitter. 01:12:39.590 --> 01:12:41.360 Oh, sorry, X the everything up. 01:12:41.540 --> 01:12:42.420 I will call it. 01:12:44.440 --> 01:12:47.120 Ping me and we'll figure out. 01:12:48.660 --> 01:12:51.100 Yeah, I think that's my message. 01:12:51.860 --> 01:12:52.580 Yeah, awesome. 01:12:52.640 --> 01:12:54.520 Well, it looks like a very ambitious project 01:12:54.880 --> 01:12:56.900 and I think you've done a lot of interesting things. 01:12:57.080 --> 01:12:58.520 So thanks for coming on and sharing it. 01:12:58.740 --> 01:13:00.800 Tell people where they can stay in touch with you as well. 01:13:01.140 --> 01:13:08.700 Yes, so my GitHub handle is G-I-0-B-A-R-O. 01:13:09.500 --> 01:13:14.120 You can find me with the same handle 01:13:14.140 --> 01:13:23.960 Twitter. I have a blog, even if I don't write a ton, but you can find my blog at blog.baro.dev. 01:13:24.380 --> 01:13:35.900 And if you want to join me and Marcelo Trilesinski, again, author of Uvicorn, Starlet, and a bunch of 01:13:36.800 --> 01:13:42.700 stuff he's making with TokenMaxing, we recently started a podcast available on YouTube. 01:13:44.440 --> 01:13:46.620 at the www pod. 01:13:47.060 --> 01:13:47.160 Nice. 01:13:47.600 --> 01:13:49.000 So people, check them out. 01:13:49.140 --> 01:13:49.680 Check out the pod. 01:13:49.840 --> 01:13:50.260 That sounds fun. 01:13:50.340 --> 01:13:51.620 I'll give it a look as well. 01:13:51.780 --> 01:13:52.920 Joe, thanks for coming on the show. 01:13:53.160 --> 01:13:53.880 Nice to catch up with you. 01:13:54.120 --> 01:13:54.840 Thank you for having me. 01:13:55.140 --> 01:13:56.400 It was super good. 01:13:57.700 --> 01:13:59.840 This has been another episode of Talk Python To Me. 01:14:00.220 --> 01:14:00.940 Thank you to our sponsors. 01:14:01.240 --> 01:14:02.420 Be sure to check out what they're offering. 01:14:02.660 --> 01:14:04.000 It really helps support the show. 01:14:04.560 --> 01:14:07.060 This episode is sponsored by Sentry's Seer. 01:14:07.380 --> 01:14:09.800 If you're tired of debugging in the dark, give Seer a try. 01:14:10.360 --> 01:14:12.500 There are plenty of AI tools that help you write code, 01:14:12.720 --> 01:14:15.580 but Sentry's seer is built to help you fix it when it breaks. 01:14:16.340 --> 01:14:20.420 Visit talkpython.fm/sentry and use the code talkpython26, 01:14:20.880 --> 01:14:24.040 all one word, no spaces, for $100 in Sentry credits. 01:14:24.740 --> 01:14:26.360 And it's brought to you by us. 01:14:26.960 --> 01:14:30.080 Talk Python and Python Bytes both now have MCP servers. 01:14:30.660 --> 01:14:33.460 Point your AI at 10 plus years of Python episodes, 01:14:34.100 --> 01:14:35.600 transcripts, and show notes free. 01:14:36.020 --> 01:14:39.800 Click MCP in the nav at talkpython.fm and at Python Bytes. 01:14:40.200 --> 01:14:42.080 If you or your team needs to learn Python, 01:14:42.380 --> 01:14:48.160 We have over 270 hours of beginner and advanced courses on topics ranging from complete beginners 01:14:48.450 --> 01:14:52.300 to async code, Flask, Django, HTMX, and even LLMs. 01:14:52.660 --> 01:14:54.860 Best of all, there's no subscription in sight. 01:14:55.400 --> 01:14:57.180 Browse the catalog at talkpython.fm. 01:14:57.860 --> 01:15:01.980 And if you're not already subscribed to the show on your favorite podcast player, what 01:15:01.980 --> 01:15:02.540 are you waiting for? 01:15:03.220 --> 01:15:04.960 Just search for Python in your podcast player. 01:15:05.120 --> 01:15:05.940 We should be right at the top. 01:15:06.380 --> 01:15:09.140 If you enjoy that geeky rap song, you can download the full track. 01:15:09.350 --> 01:15:11.260 The link is actually in your podcast blur show notes. 01:15:12.040 --> 01:15:13.420 This is your host, Michael Kennedy. 01:15:13.820 --> 01:15:14.880 Thank you so much for listening. 01:15:15.050 --> 01:15:15.880 I really appreciate it. 01:15:16.300 --> 01:15:17.020 I'll see you next time. 01:15:27.480 --> 01:15:28.420 I'm me. 01:15:29.820 --> 01:15:30.920 Get me ready to roll. 01:15:32.560 --> 01:15:33.700 Upgrade the code. 01:15:34.560 --> 01:15:36.240 No fear of getting old. 01:15:37.740 --> 01:15:39.720 We tapped into that modern vibe 01:15:40.160 --> 01:15:41.140 over Kenny's storm. 01:16:11.040 --> 01:16:11.060 Продолжение следует...