PEP 711 - Distributing Python Binaries

00:00 What if we distributed CPython, the runtime, in the same way we distribute Python packages,

00:05 as pre-built binary wheels that only need to be downloaded and unzipped to run? For starters,

00:11 that'd mean we could ship and deploy Python apps without worrying whether Python itself is

00:16 available or up-to-date on the platform. Nathaniel Smith has just proposed a PEP to do just that,

00:22 PEP 7.11, and we'll dive into it with him next. This is Talk Python to Me, episode 412,

00:29 recorded April 18th, 2023.

00:31 Welcome to Talk Python to Me, a weekly podcast on Python. This is your host, Michael Kennedy.

00:49 Follow me on Mastodon, where I'm @mkennedy, and follow the podcast using @talkpython,

00:54 both on fosstodon.org. Be careful with impersonating accounts on other instances. There are many.

01:00 Keep up with the show and listen to over seven years of past episodes at talkpython.fm.

01:06 We've started streaming most of our episodes live on YouTube. Subscribe to our YouTube channel over at

01:11 talkpython.fm/youtube to get notified about upcoming shows and be part of that episode.

01:17 This episode is brought to you by Sentry and us over at Talk Python Training. Please check out what

01:23 we're both offering during our segments. It really helps support the show. Nathaniel, welcome back to

01:29 Talk Python to Me.

01:30 How's it going?

01:31 It's going real well. It's going real well. We're on the eve of the eve of PyCon. How about that?

01:37 Eve square? Okay. Yeah.

01:40 I don't know how many eve. Maybe it's the eve to the third, but we're very near PyCon. I'm pretty

01:44 excited.

01:45 Yeah. Anti-penultimate eve.

01:48 I don't know.

01:48 The penultimate eve, perhaps? Yeah, exactly.

01:50 I suspect a lot of people will be listening to this show on their way to PyCon. So if you are,

01:55 awesome. Come say hello to me. I'm going to be doing some live shows, some Ask Me Anything,

01:59 some various other things. Are you going to be at PyCon this year?

02:01 I'm not. I'm not, unfortunately.

02:03 So they're going to have to just shoot you a message on Twitter or on Mastodon or something like that, right?

02:09 Yeah. I'm easy to find. GitHub, email, whatever. Yeah.

02:13 Cool. Well, everyone going to PyCon, hope you have a great time and do come say hi. And with that,

02:20 you know, we're going to be talking about this project, this new PEP about distributing Python

02:27 itself, kind of like you distribute Python packages, but a little bit more. Why not? I mean,

02:33 it seems pretty reasonable to me. And I'm super, super excited to see work in this area because

02:39 Python is so strong in so many areas. And there's just a couple of like really big gaps that other

02:47 technologies have nailed so well. Two of them that I see that are super significant is like,

02:52 hey, Michael, I want to build a mobile app. How do I do that in Python? Or I want to build a desktop

02:56 app. How do you do that in Python? Like, I'm not sure. I'm not sure you should even think about

03:01 doing like desktop, maybe mobile. I mean, Kivy is great, but it's not like a general purpose

03:07 UI toolkit. And so that's the one. The other is, hey, I have an application. I want to give it to

03:13 someone who is not a developer and have them run it. Right. And there are some tools that address

03:19 that. But one of them is just like, oh, how do they get Python at all? And your project, your pep,

03:24 and some of the ideas relate specifically to how do we make it easier to get a prebuilt

03:30 non-admin, not install for the whole system, Python for somebody so they can run an app or

03:37 even for developers, right? Yeah. I mean, in fact, I mean, I'm a developer, so that's kind of in some

03:43 ways the original use case, you know, scratch your initial app, right? Yeah. But yeah, I mean,

03:48 it's just, it's a very general capability, I think, once you have it. So yeah, I mean,

03:54 the motivation there is basically like, you know, there are lots and lots of ways to get Python,

03:57 right? You can get it from, you know, the Windows store, it comes with on your pre-installed on your

04:03 Mac, but not a very good one. Yeah, not quite, sort of. There's, you know, but there's also Homebrew

04:09 or Pyenv or your Linux distro has it. And you can, you know, get it through Conda. And if you download

04:16 Blender, oh, there's secretly a Python inside, you know, like, it's just, you know, there's just so

04:20 many different ways to do it. And that's great. You know, like, it's good to have all these options.

04:26 They also are different use cases, but it's sort of silly that, you know, it's obviously it's,

04:31 you know, it's flexible. There's lots of ways to do it, but there's no way to just be like, okay,

04:36 like, I just want a zip file that has Python in it. And it's like a standard way that's like supported

04:41 and, you know, kind of, we all use, so we can call sort of share benefits and improvements and all of

04:46 that. So that's kind of the key, you know, the PEP is not that innovative in terms of what it's

04:51 actually, you know, doing, right? It's a zip file with Python in it. But it's just sort of trying to

04:57 do kind of the logistics of like, okay, but like, let's all agree on how we're going to put it on PyPI.

05:01 Let's, you know, have tags and stuff. So tools can figure out what they're looking at and do stuff

05:05 automatically. And I think that unlocks a lot of use cases, just that one simple change.

05:09 I think it does as well. I mean, your abstract is one of the more concise ones, I would say.

05:15 Tell people about the abstract here.

05:19 Yeah, the abstract on the PEP is, it's, quote, like wheels, but instead of a prebuilt Python

05:26 package, it's a prebuilt Python interpreter. That is the full abstract. I figured that basically,

05:30 you know, tells people what they need to know.

05:32 Yeah. So the idea is kind of like you would say pip install request, you might say pip install

05:41 Python 3.11, except for that you can't use the word pip because pip is built on Python. And so

05:46 you need Python to run. I mean, it's a little bit circular there. So you kind of need something

05:51 outside of Python. But, you know, conceptually, it's, I have these things I need to run my app,

05:57 I need request SQLAlchemy, and beautiful soup. I also need Python 3.11. So those are my dependencies.

06:03 Give me that, right?

06:04 And I mean, you could even imagine potentially pip install Python 3.11 working, I suppose,

06:09 like, you would need a Python, some Python to run pip. But once you have that, then it's probably

06:14 still convenient to be able to say, okay, actually, shoot, so I got this bug report saying in 3.11.2

06:19 specifically, there's some issue, and I'm not sure if I can reproduce it, you know, like,

06:22 being able to just grab that in one command, pretty useful. But that said, you know, yeah, it's

06:29 not necessarily, pip isn't necessarily the target. I've been working on some stuff there as well.

06:36 I don't know if you want to get it together.

06:38 Yeah, we'll definitely get into it. We'll definitely get into it. I think,

06:41 you know, you need something a little bit on the outside. And I think ideally doesn't depend on

06:46 Python being on the system, because that it would be perfectly useful for it to depend on Python.

06:51 And this gives you a different version, this gives you a way to quickly toggle between these versions

06:56 and these different setups. But if you could omit that dependency on Python, then all of a sudden,

07:00 you give a way to give it to people who are not developers and use cases where it's not just I

07:05 already have Python and I need to zoom in. Maybe you're a developer, but you're not a Python developer.

07:09 Should you have to manage your own Python installation so that you can use, you know,

07:14 something that needs Python to run against your source code, right? That is not Python. So there's

07:19 a lot of scenarios where I think it gets unlocked if you use a different foundation.

07:24 Yeah, I mean, some of the one of the audiences definitely have in mind here is like, you know,

07:29 people with like taking their very first ever programming course on the first day of class.

07:33 Like right now, it's pretty awkward that you're like, okay, well, first, you have to go to Python.org

07:37 and then click through here and click there and download that. Oh, wait, no, not that version.

07:41 Oh, did you forget to check to put in your path? Oh, dear. Hold on.

07:44 Yeah. And oh, no, do you use the Pylauncher? Are you on Windows? Are you like, it was just,

07:48 it's this extra fiddliness. And you know, it's funny, we've spent all this effort

07:53 in the last few years kind of getting wheels to the point where they can just work, right? You can

07:58 just, you know, pip install NumPy and it works everywhere and, you know, stuff like that. But

08:01 Python itself isn't there yet. Another use case, I've sort of maybe my primary use case,

08:08 sort of audience I have in mind is, right, I develop Python packages and as open source and

08:14 distributed on PyPI, stuff like Trio. And so I have the problem of, you know, I want to be

08:21 welcoming to new contributors, right? I want to, you know, bring them in, get them started first,

08:26 quickly. They're volunteers. So like, I don't want them like faffing around and struggling and getting

08:31 stuck trying to just like run the tests or anything like that, right? Like both, that's just a waste of

08:36 their time. It's like kind of rude and inconsiderate. And it's also, you know, there's likely they'll just

08:40 give up, you know, if it's just a casual, like, you know, they aren't really invested yet. It's just

08:45 something they're doing for fun or out of interest. You know, I really want to, you know, make that easy

08:49 for them. And so part of the vision here is like, you know, be able to say, oh, yeah, so you check out

08:56 Trio, you know, here, you type this git clone command, and then you, you know, you have, you know,

09:01 this, some kind of Python management tool installed. And you type that tool, you know, run tests,

09:07 and it makes sure you get the right version of Python and set up the environment correctly. And then

09:10 it executes it for you. And it knows what tests are, how to run the tests in this project, because it

09:15 looks at, I don't know, pyproject.toml or whatever, right? And so just sort of capturing all, you know,

09:21 we have all those pieces, we don't really have anything that kind of brings them all together

09:24 into that, like, just type one thing. That's it. It's going, you know, and it just works.

09:30 We don't have many, we honestly don't have many tools that are outside looking in, in Python. So much of our

09:36 tooling and our infrastructure is you have Python. Now, now you use the tools, now you install it,

09:41 now you install black or, or rough or other. Yeah, there's just this kind of very old assumption,

09:47 which I mean, it made sense, like 10, 15, 20 years ago, where sort of everything was sort of

09:53 installed manually. And of course, you have, you're going to go through some work to like get set up

09:56 with Python. It's the foundation of your whole environment. And then, you know, kind of we add

10:00 stuff to make it easier on top of that, but kind of, I think it's time to kind of go back and re,

10:04 reevaluate that sort of foundational assumption. Yeah, absolutely. So you mentioned Trio, I know

10:11 before we dive too much further in, I'm going to give you a chance to kind of, let folks know what

10:16 you're up to. we'll talk about Trio at the end if we've got time, but you know, what have you

10:21 been up to since June 29th, 20, 2018, you know, five years ago, roughly last time I was on the podcast.

10:28 Yeah. wow. That was really early in Trio's life, I guess, actually. so, I mean, I've had

10:37 a lot of just, you know, real life, has happened. then I was, you know, sick for a while and trying

10:44 to kind of get back on my feet, did some consulting, just started a new job. So that's

10:50 kind of, you know, a lot of distractions, but, also, you know, yeah, I think, Trio is still,

10:56 you know, I still like it a lot. it's definitely had more influence. Actually,

11:02 had I even published the structure concurrency blog post then? I don't remember.

11:06 I feel like it sounds familiar to me, even though it's been five years, it does sound familiar. So I

11:11 do think so. But what has happened since then certainly is Python has seen some of these ideas and

11:18 adopted them, right? Like, like 311s, concurrency stuff. Yeah. So there's been a, like sort of the,

11:25 the influence has gone a lot further than I ever expected. both actually in other languages. So

11:29 yeah, like Swift and, Kotlin and, you know, have all kinds of adopted ideas from here. Java

11:37 apparently is making some big changes, coming up soon with a whole new concurrency setup. And they're

11:42 like saying like, yeah, we're basing it on that Nathaniel Smith random blog post. Okay. Yeah.

11:48 Okay. Okay. Yeah. It's very flattering. but yeah. and yes, also, in the Python itself,

11:59 it's sort of complicated because we, it's sort of this awkward situation where there's a sink.io that's

12:03 in the standard library. And then there's my sort of competing thing trio, which, yeah, we guess we

12:08 should say trio is an async library for Python. that's, you know, portable, it's sort of, you know,

12:13 an alternative to a sink.io. There are some tools to kind of let you use both at once, but, it's not

12:18 a library for a sink.io it's its own thing. and so, you know, like, obviously we all wish there was just

12:26 one obvious choice. I kind of looked, you know, but async.io is also in this very difficult

12:31 position being in the standard library, and being sort of built up over time. And a lot of it

12:36 was designed before we even had stuff like a sink await. so there's just a lot of machinery in

12:42 there that's kind of already committed to other ways of doing things. And it's very hard to change.

12:46 And trio was sort of like, well, look, we have all these modern things and some new ideas coming in,

12:50 like structure. Concurrency is a better way to kind of, write your right concurrent programs.

12:55 and it was able to set out a clean slate to like, really, you know, do that all from the start

13:00 and be much simpler. so that was important, you know, to have it be its own thing just so we

13:06 could, you know, work that stuff out. Then there's a question of, okay, now what do we all switch to trio?

13:09 Do we move it back into a sink.io? Do they both continue?

13:12 That's been a debate for some really popular things. I think that is interesting. A lot of

13:17 people say, well, why is library X, which everyone uses, why is that not built into Python? Why do I

13:23 need to pip install it? And a lot of times the answer is because making it part of Python will

13:28 harm its ability to innovate and change, right? It'll slow it way down.

13:32 Yeah. Like there was a whole debate some years ago, but like, you know, like we all know the

13:37 HTTP client in Python, you're a lib or whatever. It's just really bad. You should just never use

13:42 it. And it's broken a lot of ways. Like it's just, you just don't use it. but we still ship it

13:47 because it would be too disruptive to take it out. That's also why we can't change it. There's just

13:52 too much code out there depending on all the weird quirks. and we don't want to ship something else

13:58 because then it'll end up being your, like URL, you know, five years later. so we, you know,

14:03 it was questioning, like, should we put requests in the standard library or URL lib3 or what, you know,

14:07 one of these, and it's just, you know, then you can't ship security fixes. You can't improve

14:12 your API. You can't, you know, so as we've gotten better at packaging also, it's taken some of the,

14:18 pressure off the standard library to be all things to all people.

14:21 This portion of talk Python to me is brought to you by code cov from century. Have you heard

14:28 about code cov? They are the leading code coverage tool on the market, and they just joined century,

14:34 the error tracking and performance monitoring company that you know, and love. Code cov is

14:38 the all in one code coverage reporting solution for any test suite, giving developers actionable insights

14:44 to deploy reliable code with confidence. Code cov is easy to set up. If you are already both a code

14:51 cov and century user, GitHub integration is even enabled automatically for you. You'll get coverage

14:57 insights directly in your workflows. Code coverage pull request comments allow you to quickly analyze

15:02 your PR's coverage and risk without leaving your workflow. It'll reduce the guesswork. You set up

15:08 customizable quality gates and let your continuous integration do the rest. And code cov identifies where

15:15 tests can help you avoid errors in production through their century integration. If an error does occur,

15:21 you'll even see code coverage details directly in your stack traces. So you can see the untested,

15:27 partially or fully covered code that may be causing errors to help you fix your tests to avoid similar

15:33 errors happening in the future. Get started for free or take advantage of century's promo

15:38 pricing where with a century team or business plan, you can get your first five row code cov seats for

15:45 just $29 a month. That's a 40% savings. Visit talkpython.fm/century to get started. Remember to use the code

15:53 talkpython to let them know you came from us. It really does help support the show. That's talkpython.fm/s-e-n-t-r-y

16:02 century and the code talkpython. Thank you to century and code cov for supporting the show.

16:08 When the standard library first came into existence, there was no PyPI and there was no package. Like,

16:17 it had to come with it because how else you're going to hunt it down on Usenet and on, you know,

16:23 base 64 and code it and put it like, what are you going to do?

16:26 Yeah. Or I mean, maybe you find it, you know, you download, I don't know, Twisted or something like

16:30 from an FTP site. Yeah. Or an FTP site or something like that.

16:32 You have to manually unpack it and have to put it in your pipe. Like it was all totally,

16:35 yeah, you know, knocking rocks together.

16:38 Sharpen sticks.

16:41 You hope it's flint and it creates a spark.

16:43 Yeah.

16:43 Yeah.

16:52 So I think even though we're already far down the road and making changes is breaking and doesn't

16:57 make sense, it might make sense to ship less in the standard library, quite a bit less and just say,

17:03 oh, you're going to pip install some meta package that explodes out some section. I'm going to pip install

17:08 the collections area. Boom. And now I got a bunch more potentially. Who knows?

17:12 One thing I'd really like to see as a possible sort of future there is moving some of the standard

17:18 library into wheels that are installed by default.

17:21 Yes, exactly.

17:22 So you get, so, you know, that's sort of, it's sort of this halfway house, right? Where, you know,

17:26 it's still the case you download Python, install it, they're there. So we don't just like break

17:30 everyone in the world who just assumes they're there. But then it kind of gives us that both of,

17:37 you know, the long-term, if it's like, we want to get rid of it, it kind of gives us,

17:40 or push it out to PyPI or just remove it entirely. Then it gives us kind of a way to do that more

17:46 gradually. But also for, you know, libraries like asyncio that are big and complex and really,

17:51 you know, would benefit from being able to have their own release cadence and bug fixes and deprecation

17:56 cycles and all of that. Then it's like, yeah, it still ships with Python,

18:00 Python, but then you can pip upgrade it. You know, you're not stuck with that exact version.

18:04 That could only change when a whole new Python release comes out and you have to take all those

18:09 changes together at once.

18:10 Yeah, I, I've absolutely had this thought and I think it is, it's a really elegant solution because

18:16 on one hand, it lets the core developers focus more on the true essence of Python and it lets it be used

18:23 in more locations, right? Think PyScript for example, or MicroPython, right? It might be that you can

18:29 create a central core that is exactly the same on all of these. You don't have to consider like,

18:35 of course, this is what runs, it runs everywhere, but you still get that backwards compatibility and

18:40 you get the ability to say, actually, I want the newest version of asyncio because I want this

18:45 more high performance background worker or something.

18:48 Yeah. Or even just, I mean, for smaller, like, you know, I don't want the newest version of async

18:53 IO because like, I don't know if it works, but I want to install in a scratch environment, this like development

18:59 version so I can try it out and give them feedback before they, you know, really make the release and

19:04 set the API in stone. And again, like right now, like you'd have to go build your own Python and it's

19:09 like, it's just kind of a whole thing, right? You can't just do install --pre.

19:14 Right. Exactly. It's definitely a more of a barrier for people who are just casually wanting to test stuff

19:19 out. You got to be pretty committed to getting Python 312 alpha six or whatever we're at. Right. I don't

19:25 know. Yeah. Yeah. Yeah. Indeed. Okay. very cool. Well, maybe we'll come back and dive into tree a little

19:31 bit more, but yeah. And so what do you, what are you doing these days? You talked about doing a little

19:34 consulting and, well, yeah, well, so I just started at, a new job, just like last week, week before,

19:41 I guess. Yeah. It's been less than two weeks. Yeah. and working at Anthropic, which I don't know if

19:48 anyone's heard of, it's still somewhat stealthy, but, it's sort of, yeah, I mean, quickly

19:55 changing. I don't know the exact status currently. but yeah, so my understanding of the background

20:00 here is that, there's actually the team at open AI who trained GPT three, just sort of,

20:07 you know, sat down together and thought, you know, decided they really wanted to do more of like a pure

20:13 focus on like interpretability and safety. Like, how do you get these models? Like, how do you know what

20:18 these models are actually going to do and how do you get them to do what you want instead of kind of,

20:22 you know, stuff like making things up or just, you know, we've all kind of got the scene now how

20:27 these large language models can go just all over the place, do all kinds of strange things.

20:33 and so, and they decided to be in suit for slander, I believe. yes. Yeah.

20:38 Somebody in the UK, I think. Yeah. Well, yeah, there's definitely, there's one that just like,

20:43 if you ask the model, like, you know, can you give me some advantages of like, you know,

20:48 problems with sexual harassment in law schools and it just picks like five real law professors

20:52 and makes up stuff like it's really, really bad. Yeah. And it like sites sources that are all made

20:59 up. Like it's, you know, like they, you know, they're just, they're very powerful, but also

21:04 not well understood or how to like kind of make them useful and safe.

21:07 Just a little bit of devil's advocate though. They are incredibly powerful and they are incredibly

21:13 capable. And that's, I think part of the dangers you're like, oh my God, it knows this. Oh my gosh.

21:17 Yeah. It understood all of that. And I ask it. So the fifth thing it says, you're like, well,

21:23 at this point I'm convinced that it really is on. And then maybe that's the made up one. And I think

21:28 that's the dangers. Cause it's actually, it's almost an incanny valley. It's close enough to write that

21:33 you're like, okay, this thing's right. It knows. Yeah. So yeah. So personally, like I'm still kind

21:37 of up in the air on how you like, how impactful they'll be, where the impact will be. Like,

21:42 I think it's just a lot of open questions. You haven't bought a farm, like a goat farm in the

21:47 woods. Cause you just give them technologies. Okay. But I guess I do have stock options now,

21:56 apparently, or I will at some point if they vest. So I guess that's the other route. But no, but anyway,

22:02 so I was just saying, so yeah. So, Anthropix is just interesting company where, you know,

22:08 you actually get to play with some of those big models internally. They're kind of working on

22:12 on releasing products now. but it's also been kind of just a really interesting to kind of

22:19 get the sense internally of like, it's really kind of this like research culture, which is appealing

22:24 to me with, I'm sort of coming out of academia of a lot of like numerics background. and what's

22:29 also interesting is that part of the reason, we kind of, you know, connected is that apparently,

22:35 you know, turns out a ton of their internal infrastructure runs on trio. so, they're,

22:40 you know, partly hired me to support that, and are actually giving me time to work on open source,

22:45 paid time. so actually they are, funding this PEP 7.11, you know, Python binary stuff, though. They don't know it yet. now they do. Yeah. Yeah. Yeah.

22:56 If they listen to the podcast, they'll learn on this. That's great. That's really, that's really cool.

22:59 It looks like a, an interesting area to be working. I agree that the research oriented places,

23:04 they are fun area to work, right? You're not. Yeah. And there's just a lot of flexibility,

23:08 kind of, you know, like it's clear this stuff is going to have effects, which effects and how big

23:14 and all that. I don't know, but you know, being at ground zeros, you know, it's exciting. It's really

23:19 exciting. Yeah. And a lot of chances maybe have some impact. So cool. All right. Let's dive into the PIP,

23:24 the PIP, the PEP, the PEP. That is not, it's not quite PIP, but it's kind of like PIP. Okay.

23:28 Yeah. Okay. So we talked a little bit about the motivation. we talked a little bit about what

23:35 it is. maybe, maybe tell us a bit about the spec. Like what is the PEP actually say?

23:42 what is it actually trying to deliver? And we can talk about like the use cases and

23:45 Yeah. the tools for it and so on later. Yeah. I mean, and so, like I said, you know,

23:51 that the abstract deals like wheels, but it's an interpreter instead of a package.

23:56 that's partly just sort of a tagline of like how you use it, but it's actually also a lot about how

24:01 the actual spec is written. it's just sort of like, well, you know, we, we've done a ton of

24:06 work over the last, you know, five, 10 years. A lot of people have put a lot of work into making wheels

24:11 work, right. In terms of like, figure out, okay, how do we, you know, have metadata that's usable to

24:16 like keep track of, you know, which packages are installed and their versions and which ones are compatible.

24:21 And if for a binary build, like which systems can you put this on and, you know, all the many

24:26 Linux work and, just all of that stuff. And it's just like, well, you know, so we have wheels,

24:32 we don't need to reinvent the wheel again. so I'm just sort of taking all of that.

24:39 so it's just like, okay, it was mostly, it's defined, it's just a Delta against the wheel spec. It's like,

24:44 okay, and the wheel spec, you have, you know, this directory for metadata, I have, you know,

24:49 that same directory, but I call this calling these Python binaries, pybys, just to have sort of a

24:54 short name, you can stick it in a prefix or sorry, in a file extension.

24:58 Yeah. P Y B I. Yes. Dot P Y B I. I like it. Yeah. The PYPY, the interpreter and PYPI, the,

25:06 package, repository work confusing enough. So I thought I'd add another near homonym to the,

25:11 PYPY. It's PYBI. Yeah. Yeah. You know, anyway. but so, yeah, so like, but they look like,

25:20 you know, the file names look like wheels, like, you know, be something like, you know,

25:24 cPython dash version dash many Linux to 17.py bi. they context looks like wheels. They're

25:31 basically just zip files. There's some, you know, instead of a .dist info directory, you have a

25:36 pybyinfo directory and it has a metadata file that's in the same format as wheel metadata files with,

25:41 for the name and version and, you know, description, all that stuff.

25:46 there, there are a few tweaks, basically just what, you know, you need specifically for,

25:52 interpreters. so, okay. So like one thing that makes it a lot simpler is that, for an,

25:59 there's only one interpreter in a Python environment, right. Whereas wheels are kind of designed to be

26:03 flexible and be installed into different kinds of different kinds of Python environments with

26:07 different layouts. a PYBI, it's just like, it's just a raw set of files. You unzip it. That's it.

26:14 You're done. where wheels, there's like, well, okay, if you, you want to put this in site

26:17 packages, so you have to go find that. Whereas this goes in the bin directory. So you have to go

26:21 find that and do the special, you know, so that part's just, you know, not relevant. we leave

26:25 that out. there's some, slight, you know, we have to support sim links, which wheels don't,

26:33 mostly just cause there's never been a big compelling reason. What's that?

26:37 Or the windows folks out there, maybe, and others who, Oh yeah.

26:40 Yeah. What the heck are sim links?

26:41 Yeah. Okay. Well, so yeah. Sim link is a classic Unix concept, though. Windows does have them too.

26:46 Now, I guess where it's like a special magic file that, instead of having like its actual contents,

26:53 it just lists says, go look at this other location on the file system for my content.

26:57 Right. It's like, like an app shortcut, but yeah.

27:00 But like programs, not for UI.

27:02 Yeah. Well, like built into the file system.

27:04 Yes, exactly. So you try to open it, it goes to the other one.

27:07 Yeah. The, the, the operating system automatically opens that other file for you. but you could

27:11 also like look at, you know, if you, you can like say like, can you show me the sim link? And like,

27:16 it'll tell you about it if you ask, but if you don't, then it just, you know, magically works.

27:21 And it's mostly, it's just, it turns out that traditionally, Unix pythons tend to use these,

27:27 both for things like, you'll, you know, in the, your bin directory, you'll have the Python

27:32 executable. And then you'll also have Python three as a sim link to Python and Python 3.11 as a sim

27:37 link to Python three. And so, you know, one to preserve that. and also it turns out on macOS,

27:43 they have this very specific kind of layout they want with like framework. Like,

27:47 I don't really understand it in detail, but like there's sort of like a, how a macOS app is

27:51 supposed to be structured and that it turns out to involve sim links. So we just, you know, we just

27:55 have to support, that said, I mean, the way we support them is like, it turns out there's a

27:59 standard way to put them in zips. So I say, let's do that. You know, like again, really trying to keep

28:04 this as boring as possible, you know? I did know. and then the last thing file, that's crazy.

28:09 Yeah. It's a, it's true. It's an extension from the info zip folks, but then it's become,

28:14 I don't know. Zip's a strange format. It's kind of like an oral tradition as much as like an actual like specified format, but there's an entire documentary on zip.

28:24 And, I believe I came up with it. There's yeah, it's even controversial in its early days.

28:30 It's it's nuts, but yeah, it's even won the compression de facto standard these days for the

28:35 most. Yeah. It's definitely got trade-offs, but it's just in terms of, it's just really useful.

28:39 It's just a thing that everyone, everything could understand. It's just, it's just so compatible and it's

28:45 also convenient that you can do random access, unlike some of the alternatives. You can just,

28:50 you can pull out one file from the middle if you want to.

28:52 The fact that anyone can open it is so much better than it might save one more percent. Yeah, for sure.

28:57 Yeah. Cool. Okay. So we've got these, basically the PI by file is the zip file. Is that basically the

29:08 entire interpreter just kind of bundled into a zip file? Like what's the deal there?

29:13 Yeah. Okay. I mean, it's just literally like, you know, you install Python into a certain directory

29:18 and then you take that directory and you put it in a zip file. There's a little bit of tweaking to like

29:22 make sure it's self-contained and you can move it in a portable, portable, relocatable, I guess is a

29:26 better word. Yeah.

29:27 Yeah. So sometimes if you just install Python regularly, it's kind of, we'll have hard coded.

29:32 I know that I'm at this particular position on the file system. And so I need to make sure we don't do

29:37 that. And also to make it self-contained, it's like the same thing we do with wheels. Like you have to

29:42 vendor some libraries, right? If it wants to use read line as a library to, for like in the REPL to be able to

29:49 like edit your line as you're typing it, then, you know, we can't just assume it's on the system.

29:53 We have to include that inside the PyBuy. So, and, but again, like this is stuff we've all

29:59 already dealt with with wheels. There's tools for doing it. We understand how to do it. And I'm just

30:03 reusing those tools. So we're not, if I were to run a Python application delivered by one of these PyBuy's,

30:12 does it have to unzip the contents into a location and then run it there? Or can it just run it

30:17 straight out of memory or what does that work?

30:20 Well, so by itself, the format, I mean, it's just a zip file, right? So you can do with it,

30:27 what you can do with a zip file, which I mean, is not much on its own. You need some software to work

30:30 with it. Right. Now that said, I think, so, yeah. So like, if you just were starting with nothing and

30:37 you're like, I just, I have a URL to some PyBuy and I want to use it, then you'd have to download it,

30:42 run it on zip tool, and then you'd have a, you could go into that directory. It's a Python environment.

30:46 You know, you could run pip in there or whatever. That said, I think this is a really useful building

30:52 block for tools that want to go beyond that. So things like delivering a prebuilt application that

30:56 you can just run without unpacking. Like there are various tools to do that, like PyOxidizer,

31:02 PyToApp. I don't know. There's a ton of them actually. I'm probably forgetting like 10 more.

31:06 Yeah. The ones that come to mind for me are PyToApp, PyInstaller, and PyOxidizer for sure.

31:12 Yeah.

31:12 PyOxidizer being the newest of them.

31:15 Yes. Oxidizer because it involves Rust somehow.

31:18 Because all the things that involve Rust.

31:21 Yes. So, but yeah, so those tools that it's really useful to be able to say, okay,

31:29 like I'm going to do some clever thing to like set up, I don't know, a self-extracting executable or

31:34 whatever it is they do for their distribution mechanism. I'm going to create an installer

31:38 program. I'm going to, whatever it is, but you still need an actual Python to put into that.

31:43 Right. And so having a straightforward way where it's okay, that's not their problem anymore to

31:48 figure out how to find a Python and get it built and working for the target system.

31:52 So they can just say, okay, there's, I can just like grab, you know, okay, yeah, you want to target,

31:56 you know, many Linux. Cool. I'll just go grab the right Python. It's already there. I know it works.

32:01 And now I can take the files out of this PyBI and do whatever I want with them. I can pack them

32:06 into my installer or do clever things to make them usable out of memory or whatever. And they can focus

32:13 on that part instead of just the, like, how do you even get a Python? Yeah. Or how do you once

32:20 find yourself in the wrong Python, get the right Python. Yeah. Yeah. Yeah. That's a, I don't know if

32:26 that's trickier or less tricky, right? It's one thing to say, dear user, go get Python. You need

32:31 that. It's another thing to say, go upgrade your Python and hope you don't break something.

32:35 You know, I think. Yeah. Well, but also that's part of the point of these being self-contained is,

32:40 so, I mean, this is one of the more trivial use cases, right? But right now we all use virtual

32:43 ends and mostly that's fine, but also sometimes, you know, they can get, you know, janking stuff

32:48 could happen. Like, you know, you're on Linux, you do an app to upgrade and now your system

32:52 Python's changed and all the virtual ends that were based on it are broken now.

32:55 Yeah. Yeah.

32:56 Because it like had some kind of dependence on that exact binary. now I will say you

33:00 would always want to do this, but at least it's nice to have the option. You could say,

33:03 okay, instead of making virtual ends, I'm just going to make real ends. They're all just going to be,

33:07 I'm just going to drop a new copy of Python in each environment. And that way I just,

33:10 it's totally self-contained. I know exactly what I have. It upgrades when I decide to upgrade it.

33:14 and it's just, you know, it's a nice, it's a nice option, right? Sometimes to have that.

33:18 and, and also, you know, it gives you that total isolation, right? So you, you, then we were

33:23 just saying about that issue of like, oh, I wanted to use this. So I went and installed, upgraded my

33:27 Python. But now that other thing I was already using broke, cause they're using the same Python.

33:30 It's like very easy to say, nah, just give them different pythons. You know,

33:33 There's not that much that changes over time. That's a backwards breaking sort of thing. I mean,

33:38 two to three, but I think that's kind of, let's put that in the past. But that's a whole idea.

33:42 I did. Well, but I did recently, I was working with MongoDB using Beanie, which was using

33:47 motor, which was using the at async or at co-routine decorator, which was removed in 3.11 or 3.10,

33:56 one of those recent upgrades. And it had been deprecated forever. The people at MongoDB said,

34:02 we don't care. We're just going to leave it. Who wants to put the word async in front of my method?

34:06 That's tricky. I mean, they just probably won't pay attention. And my, my code wouldn't work. I'm

34:10 like, why doesn't this work? Oh, the thing I depend on, which the thing it depends on that,

34:14 that thing needed less than 3.10 or whatever, 3.10 or 3.11, whatever.

34:19 Yeah. Now we're back to it. I think I have struggles to adapt without breaking. Yeah.

34:25 Yeah. But yeah, stuff like this would, stuff, it does happen, you know, and this kind of isolation gives you 100% confidence to say, I'm going to make this new app. We're going to

34:33 try running this app on, on this in production, and it's not going to hurt anything. And I don't

34:38 need Docker. And you can say, or you could use, say, I'm going to, you know, use this exact point

34:43 version on in development. And then I'm going to take that and build it, create my, use that to create

34:49 my Docker image. Like I don't need the like prebuilt Docker stuff. I could just grab Python

34:53 from PyPI. And I know it's the exact same version. Everyone else is using built by the python.org folks.

34:59 Hopefully, you know, we're not there yet, but like, that's kind of the, where we're trying to

35:02 a PEP and not something on GitHub, right? Yeah, sure. Well, it has been on GitHub for a while,

35:08 but you know, I have time. Yeah. Yeah. So maybe it's, it's worth jumping over that. But before we do

35:13 two questions, maybe two top level questions anyway. So this is about the concept of kind of like pip

35:21 install Python 3.11 or 10 beta two or whatever, whatever it is. Yeah. Yeah. Does that, is there a way to say,

35:28 say, and these three packages off of PyPI? Like, can I take and kind of bring a virtual environment

35:35 effectively along with me with what you're doing so far? So PyPI is again, by themselves, I mean,

35:40 it's just an archive format, right? In a package format. It doesn't do anything that said,

35:46 obviously, yeah, part of what we want is for these to be useful for things like, you know, building

35:51 environments that have other packages in them and stuff. so that's the one other thing I didn't,

35:55 I forgot to mention about, defining the format, probably the most interesting part actually,

36:01 is that we do add some new, static metadata that we put into the package and kind of the motivation

36:07 there is that, we're trying, I try to figure out, okay, what do I need to know in order to install

36:13 wheels into this Python without running it? Right. Because right now, right. Like pip assumes that it's

36:19 running on the Python it's installing into. So anytime it wonders like, you know, okay,

36:23 like what ABIs does this Python support and what version is it? What platform am I on? It can just

36:28 ask the interpreter it's running on. Right. And it's like, okay, well, it would be really nice if you

36:33 didn't have to do that both for like efficiency. Like you want to be able to, you know, figure out which,

36:39 you know, have your like installer, your resolver, figure out which versions of everything it wants,

36:44 without having to like download and run multiple versions of Python and stuff. Like

36:48 you'd really like to avoid that. and it's also things like, I want to build a cross, I want to

36:53 build, you know, release, distributions for, you know, macOS, but I'm on windows or vice versa.

37:00 or I just want to, you know, I, I'm developing on, I'm developing my package like trio on

37:05 Linux personally, but I would like it that when I lock my, you know, version. So I know all my,

37:10 you know, collaborators are using the same versions that we figure out locks that also

37:15 work on windows and macOS. And I can't just trivial run that all those pythons, you know,

37:20 from one resolver, cause it's not running on all three at once. and we, you know, in Python

37:26 packaging does have the ability to like have different dependencies on different OSs and,

37:31 you know, it gets kind of, you know, can get very kind of complicated to figure out like,

37:35 which packages do I need where? and so I want to put a bunch of metadata into the PyBI,

37:40 like all the stuff you need to solve those problems. so yeah, so the PyBI itself,

37:45 I think normally they won't ship with any packages, maybe again, call back, maybe in the future,

37:50 we'll start moving some of the standard library into wheels that are pre-installed. You can do that.

37:54 but, I'm guessing like, you know, for now it'll, they'll mostly just be, you know,

38:00 a plain vanilla Python install, but then you could take that, you could take some wheels,

38:05 bundle them all together into a new archive if you want, or yeah, again, whatever you want to do with

38:10 it, stick it in a Docker image, whatever. It's a step towards, but not necessarily trying to propose an

38:16 entire solution of here is the interpreter and all the dependencies and the code and just run it as if it

38:23 was, it had no, no dependency on your system, just treat it as like a dot exe or a dot app.

38:29 I can just double click. Yeah. I mean, it makes that a lot easier than it is. Right. Right. Like

38:34 right now, first step is just to figure out like, how do I even build a Python that'll work like that?

38:38 And that is like some arcane dark knowledge written on a tome in, you know, black ink on black paper and

38:45 a black tomb you have to go find or something like, you know, it's just, yeah, it's not easy. and so

38:53 just having the ability to say like, yeah, just grab this file, unzip it, drop some wheels in it,

38:58 zip it up again. Now that's a package you can drop, you can hand to someone and it'll work on their

39:02 system. You know, like that's, that makes it a lot more accessible. it's not the thing I'm most

39:07 personally, like I'm not immediately going to go build that one last extra tool, but I bet someone will.

39:11 Yeah. I can imagine someone will, for sure.

39:17 This portion of Talk Python To Me is brought to you by us over at Talk Python Training with our courses.

39:22 And I want to tell you about a brand new one that I'm super excited about. Python web apps that fly

39:28 with CDNs. If you have a Python web app, you want it to go super fast. Static resources turn out to be

39:35 a huge portion of that equation. Leveraging a CDN could save you up to 75% of your server load and make

39:42 your app way faster for users. And this course is a step-by-step guide on how to do it. And using a

39:49 CDN to make your Python apps faster is way easier than you think. So if you've got a Python web app

39:54 and you would like to have it scaled out globally, if you'd like to have your users have a much better

40:00 experience and maybe even save some money on server hosting and bandwidth, check out this course over at

40:06 talkpython.fm/courses. It'll be right up there at the top. And of course, the link will be in your

40:11 show notes. Thank you to everyone who's taken one of our courses. It really helps support the podcast.

40:16 Now back to the show.

40:17 My next question is not on our shared screen here, but is what impact do you think this would have

40:26 on PyPI? First of all, do you see PyPI the way the CDN that delivers packages like Trio and wheels

40:35 like Trio? Do you see that as the same channel through which CPython 3.11 is delivered?

40:40 Yeah. I mean, so I would like these to literally be like you go to PyPI/project/cpython and it says,

40:47 like, here's the latest release and you click on downloads and it shows you the, yeah, I'd like it

40:52 to just literally be stuff you upload to PyPI. Right. And when you pip and solve from there,

40:56 it figures out the platform to pick from and it downloads that wheel and off it goes. Right.

41:00 Do you think that that would add like a huge burden to the amount of traffic or do you think it would be okay?

41:05 No. I mean, we'll have to see and adapt, but Python itself, it's like, shoot, it's tens of megabytes.

41:16 Okay. So there's plenty of other packages. There are a lot of much bigger, you know,

41:20 like go look at TensorFlow or something. There are hundreds of megabyte packages on Python that are

41:25 very popular. Also, I mean, Python.org downloads go through the same CDN anyway. It's just sort of

41:31 different infrastructure on the backend, but it's still fastly serving it and donating the bandwidth.

41:35 So, you know, in that regard, I wouldn't expect much change. And also just, you know, like people

41:41 tend to install, you know, wheels more often than they install Python. Right. Again, it's hard to know

41:46 sort of the second order effects. Like maybe virtual lens will be less popular in favor of full lens

41:51 if this takes off. And then people would start installing Python more than they do now, but still,

41:57 nonetheless, I don't think it's, you know, a huge, I don't, I wouldn't, I wouldn't anticipate it being a

42:02 huge change. And if it, you know, it turns out to be a problem that, you know, we can kind of figure,

42:08 address it then, you know. Well, you could also do to a large degree, you could do things like

42:14 pip does already that caches, you could just cache the CPython wheel, the PI BI into the user profile.

42:21 And the second, third, fourth time you get it is really the CI systems and all the doctors and all

42:26 that stuff that don't understand what a cache is or any of those things. Yeah. But then, you know,

42:31 so like, if it comes to real problems, then you go to GitHub and you're like, hey, can we work

42:34 something out so that, you know, you stick a cache in front of PI PI and, you know, like stuff like

42:39 that, like, it's not, you know, trivial, but, you know, you could talk to people and solve problems.

42:44 Like, I don't, certainly, I don't think we should, you know, hold back the entire design of like how we

42:49 distribute Python and make it available because, oh, maybe it'll be too easy and people will use it too much.

42:54 You know, like, that's a good problem to have, right?

42:58 Yes, exactly. Look, they're using it. This is terrible.

43:00 Yeah. Like first you make it easy and then figure out how to solve any problems that causes don't.

43:05 Yeah. I think we've more than once solved the problem of, oh my gosh, they're using it like

43:09 Google, Netflix, you name it, you know. Think of the benefit that you'll be doing for all the

43:14 developers, especially those who have Python skills and are looking for a job. I mean, if the popularity of

43:20 Python by downloads is one thing, if you could like 4x that, we'd all be more demand. Like,

43:25 look, really, really downloaded now.

43:28 Right. Yeah. Just go out there and just download it five times in every CI job. Just, you know.

43:33 Exactly.

43:33 Just throw a forum away, but you know.

43:35 Just do it a couple of times. Just show.

43:38 Right. Yeah.

43:39 Awesome. The question that you put into the PEP here on the screen though, is why not just Conda? And I,

43:46 not being a particularly data focused person, I definitely prefer using pip over Conda because

43:52 especially it seems like a lot of the web packages are not as close to update, up to date. You know,

43:58 there's a latency before it hits Conda and it's like immediately on PIP. That said, there's a bunch

44:02 of people who are like, I kind of use Conda for this.

44:05 yeah. And, and right. If you're just like, look, I don't really care about all this. Like,

44:10 I just, you know, I want to run my Jupyter notebooks and I, you know, just need a Python that can do that.

44:16 And maybe, you know, some NumPy or whatever. Conda solves that really well. And this thing could,

44:22 you know, I'm working on, could also potentially solve that really well. And so it feels duplicative,

44:27 to those people and to them, it is, you know, it doesn't really, you know,

44:31 they're both two solutions that work, but there isn't necessarily reason for them to choose one or the

44:36 other. but this could also be a foundation for the way that Conda provides Python to itself.

44:44 Maybe, I don't know. Like there's a whole other question about how like we could bring Conda and,

44:50 you know, PyPI pip, that kind of world closer together and interoperate better. But it's sort of,

44:55 that's a whole like, you know, can of worms, lots of complicated stuff. I don't think,

45:00 I don't think, you know, this PEP itself is going to, be the thing that makes a big difference

45:05 there. Okay. But it's not an anti-Conda type of thing. No, no. Yeah. Well, and so,

45:10 right. And so, I mean, you can also get to see a version of this in the past, but,

45:13 like basically the way I think about it is that the key reason why we just like,

45:18 why PyPI is a critical piece of infrastructure that, you know, cannot be replaced by anything else

45:23 is not because it's, of its use for end users. I mean, it's great that end users use it and find it

45:32 helpful and all that, but like, that isn't the people who absolutely need it and could not have

45:36 any replacement. the reason why we just absolutely need it is for package developers,

45:41 because the way, again, you're talking about all those different ways you can get Python,

45:45 right? And there's all these different ways Python packages get distributed, right?

45:49 You can brew install packages. There's versions, you know, NumPy, a patch version of NumPy used to be

45:55 part of the standard macOS install. Maybe it still is. I don't know. you know, like when you

46:00 install Blender, there are Python packages in there, install some game using, was it RenPy? it's

46:08 going to have Python packages in there or just, you know, there's just like, there's so many different

46:12 ways that Python code goes out of the world and gets used in all these different contexts.

46:17 and if you're developing some upstream library, like, you know, trio again, or, but, you know,

46:22 or requests or NumPy or anything, then what you absolutely don't want to do is have to maintain

46:29 a separate distribution for all of those different things. You don't want to have to upload your

46:33 package to conda forge and also to Debian and also to Fedora and also to Blender forge or like,

46:41 like you just like, that's not, that doesn't make any sense. Right. so we need this, you know,

46:47 and then having every different package maintainer do that, like that just would be terrible, right? We

46:52 just would be unworkable. so the critical role that PyPI serves that just nothing else can

46:59 is it's this intermediation point between package uploaders and package users and including package

47:05 redistributors. and so, and so, you know, like I, you know, make a release of my package,

47:10 I upload it to PyPI and then that's where conda forge gets it. That's where Debian gets it. That's

47:14 where, you know, end users get it. If they're pulling straight from PyPI, like sort of it's fans out from

47:19 there. And the key difference in terms of design between pip and conda is that pip's metadata formats

47:26 and you know, wheels and the metadata and sourcetists and all that are designed around this abstraction

47:32 of, you have some kind of Python environment, but it could be any of those, right? It could be on

47:37 different OSs. It could be different, you know, ways of building it, different layout, different

47:41 pieces could be missing, like whatever you could be laid out in all kinds of different ways. I just know

47:46 that there is some kind of Python environment and I have the metadata to like figure out how to adapt to

47:52 how this particular Python environment is put together. And conda on the other hand is one of

47:58 these sort of downstream systems. It's, it can, the reason people like love it and like data science,

48:02 right, is because it's a full fledged, like arbitrary application distribution thing, right? You can install

48:08 random, you know, C libraries and you know, you can install R and R packages. Like, like it's just,

48:13 it's got, you know, compilers that are all there in the one thing. But because of that,

48:17 it isn't have this abstraction of, oh, I can handle any Python environment. A conda package of a Python package

48:25 is set up to install in a conda Python that's laid out in the way a conda environment is laid out in the way

48:30 using the libraries a conda library has, right? And so it isn't have that flexibility. If you just release

48:36 something for conda, then it's great for conda, but it's not, you know, usable to Debian and Homebrew and all of those other folks.

48:43 And so like, sort of that's the key thing that PyPI does, right? It's like, it has that abstraction

48:48 that lets you, you kind of have the Python packaging ecosystem of all those packages and their dependencies

48:54 on each other. And then you kind of project it down into each of these more specific, specialized

48:59 packaging systems. Right. And then also, because, you know, as a, that's the, that's the other thing,

49:05 as a package maintainer, I don't just write my package and upload it. Like I'm also using all by other,

49:11 you know, all the other open source maintainers work as I do it, we're all working together. Right.

49:15 And I'm depending on their work and they're depending on mine. And so I need to be able to say like,

49:19 okay, you know, my package needs those three other packages and here are the versions that I need to be able to

49:24 download, you know, create an environment with those versions and test it before I upload my package to PyPI.

49:29 And so again, all that work has to happen at that higher abstraction level. Like you can't just say,

49:35 I'm going to take the latest version from Conda and test against it. Because that's not necessarily

49:38 the version that other packet, you know, other people will get where you take the versions from PyPI.

49:43 Those are like the original ones. I can get exactly, I can have access to anything anyone has ever uploaded

49:48 as soon as they upload it and I can test them all together. And then, you know, if Conda wants to take

49:53 some curated subset of those or whatever, that's great. That's a really valuable service, but you know,

49:59 they kind of need that underlying set of packages to curate. And that's what PyPI is for. Yeah.

50:04 PyPI is kind of the definitive source of truth as the package creator intended it to be.

50:11 Yeah. And then, so, right. And then of course, yeah, for a lot of end users, it turns out they're

50:16 just going straight to that without any intermediary works great for them. And that's really cool.

50:21 But also, you know, it's not like I don't have anything against people who prefer to go through

50:26 Debian or Conda or whatever. I think that's also great, you know, if that works better for you.

50:31 But for the folks who are, you know, developing, you know, packages to upload or who just, you know,

50:37 would rather just, you know, get stuff straight from the source, the PyBIs, I think, can solve a lot

50:42 of problems that Conda, you know, just, it just doesn't address those. It has a different focus.

50:46 Trying to make it a swap over to do that might kill a little bit of what it's, it's good for, you know?

50:54 Yeah. All right. Let's, let's see. So let's move on to your announcement here. I think.

51:00 Okay. Right. Yeah.

51:01 So over on discuss.python.org, when was this? This was January 21st.

51:06 So a few months ago, you announced PyBI and Posi.

51:11 Yes. And Posi is, we talked about this mythical pip that could pip install CPython 3.11.

51:18 Posi is that mythical pip, right? Yeah. In a sense.

51:22 Yeah. So yeah, the pip 7.11, the PyBI stuff is, is just the one, one brick in my master plan.

51:27 Oh. So, right.

51:29 Right. Because yeah, because sort of the, this vision I had in mind, I kind of alluded to earlier

51:34 talking about like, you know, okay, if somebody does, you know, get clone my project, I want them to,

51:38 you'll just run the tests and know that they have the right version of Python and the right version

51:42 of the dependencies. And it just kind of, you know, and they know how to run the, you know,

51:45 just, just do it. Right. And code all that information somewhere. And Posi is sort of my

51:51 experimental, it's not ready to use, but it does have a lot of, a lot of stuff working,

51:56 is my attempt to solve that part of the problem. So the vision is Posi is a, it is a full

52:04 re-implementation of pip with, you know, the like metadata parsers and result dependency resolvers and

52:12 archive installers and all of that. Except I rewrote it all in Rust as is the style. And, but

52:21 it also, it's all built around PyBIs. So it is, it doesn't, I mean, maybe at some point,

52:26 we'll also start supporting, you know, VNs or your user installed pythons, but, you know, for now,

52:32 sort of the, for the MVP, it's just says like, okay, yeah, you have a PyBuy, I will grab that. I will

52:37 grab packages that are compatible with it. I will arrange them all to run together based on, you know,

52:42 you just, you just say what you need. I turn that into like a lock file. I fetch those packages. I run,

52:48 you know, your test script or whatever. And yeah, I mean, that, that, that is the core idea.

52:54 Yeah. And one advantage of being in Rust is that it's, you know, just because just, so obviously,

53:00 you know, if you want to hack on it, then you need like a Rust compiler and stuff. But if you just want

53:05 to use it, then we can, you know, just take a, we can compile it down to a single binary that you just,

53:10 you know, upload to wherever, install it from wherever, and you just, you drop it on your system,

53:15 you run it and it's self-contained. It can handle everything from there. So the, again,

53:21 thinking of that target audience of like beginners, right? You say, okay, install this one program.

53:27 And now you type, you know, posy run and oh, look, you're in a REPL. And it's like, and now internally,

53:32 of course, I had to go find the latest version of Python and grab it and figure out if, which build is

53:36 right for your system and do that. But like, but you know, you don't have to think about that. You just hit

53:40 enter and it happens and there's your REPL. Or then you say, you know, posy add requests and then, or posy add

53:46 Jupyter and then posy run notebook, you know, like, and it kind of is handling the, the environments behind the

53:54 scenes. Yeah. Yeah. We did a panel discussion with a bunch of core developers around packaging recently. And a

54:04 lot of them were saying things like, I don't really want to put words in the mouth, but kind of getting the, the

54:10 sense that like, okay, so we have a bunch of tools that are really neat that live within Python, you

54:15 know, I'm thinking hatch poetry, those that, that category of tools, pip itself, even, and some of the

54:22 challenges or problems that they would like to solve, they could unlock a simpler API if it was turned

54:29 inside out, right? If the tool itself controlled Python and didn't depend on Python to get started,

54:34 they mentioned rust up as a way to get started, which is a way to kind of install

54:40 a version of rust and get started. Right. And it feels to me like this is pretty close to that.

54:45 Yeah. There's a lot of overlap for sure in terms of sort of the goals and approach and all of that.

54:51 Mm-hmm. One challenge I see is so like, for example, to run the application for the,

54:59 with the Python that's bundled up inside of one of these PyBIs is you would say posy run or posy,

55:06 some kind of file or something like that, right? Sure. Yeah. Whatever. Exactly. Yeah. Yeah. Yeah.

55:10 Yeah. Yeah. Yeah. Yeah. Whatever the CLI that's yet to be fully spec'd out comes out to be, but you

55:15 know, could you do things like, could you create, you know, speaking of sim links and other types of

55:20 stuff, could you create just in the same folder where that app lives, a Python that actually just calls

55:27 posey the Python inside instead of Python itself and, and pip that says, you know, posey run pip inside

55:34 this Python to kind of bridge, to unify the API from where people are coming from, to kind of

55:40 expose the, the same tools that are inside a little bit, you know what I mean?

55:44 Well, so yes. I mean, so the way I am currently, sort of in my current prototype, basically,

55:51 it doesn't work like that just because it felt sort of more complicated to then like try to expose

55:59 those things. So, sort of the sense is like, you know, let's see how far we can get with treating

56:05 sort of the UI paradigm of like posey is just your front end to Python. Like you don't, you just,

56:11 you start your command with posey and that's that you, that's the only command you need to know kind

56:15 of. and so, and that also allows some interesting things. So like the way posey

56:20 does environments right now is it doesn't, you can have multiple environments within a project.

56:25 Like if you need to test against multiple Python versions or you need different installs for,

56:29 I don't know, tests and for building your docs and whatever. but it doesn't actually materialize

56:36 all those as separate independent virtual environments. Instead, what it does is it,

56:41 for like each unique wheel or pyby that it needs, it unpacks that into its own directory. And then on

56:47 the fly, it assembles environments by like setting up environment variables so that it can launch a

56:52 Python in such a way that it's, you know, it picks the right Python and launches in such a way that it

56:56 sees the right packages it's supposed to see. But there's only one copy of those packages on

57:00 desk. If you have multiple environments, if you want to like try out, you know,

57:05 different versions or whatever, it can just sort of like do that, without having to go

57:10 rearrange everything on desk. and that's just, you know, it's convenient. It's just a really nice

57:15 way to work with sort of having declarative Python environments. So you never update an environment

57:20 in place. There is no environment in place. It's just on each command you run, it sort of knows

57:24 declaratively, okay, these are which packages are supposed to be there. I'll give you those packages.

57:28 So there isn't even a concept exactly of like pip upgrade or pip install.

57:32 you can just say that next time I invoke environment, I'm going to give you a different

57:36 specification for which versions I want and it'll make sure that happens.

57:41 It feels a little like Docker, right? Like if you create a Docker image and you run a container,

57:46 you want to make changes to it. You don't log into the container typically and

57:49 in that sense. Right. Yeah. Yeah. But you would just say, okay, well, we changed the Docker fire.

57:54 We shut it down and we started back up with the new up better version of itself. Right.

57:58 That, that sounds like Docker. Yeah. The main difference, the big, a big difference would be

58:02 that your Docker, then when you build it, like the actual Docker file is this big old imperative,

58:07 go scribble here and then delete that and then put something else, you know, that kind of thing.

58:10 It's not just like, here's the list of things you need.

58:13 well, it feels to me like maybe a better solution than what Docker is giving you. If what

58:18 you really just want to do is run a Python and isolated Python thing repeatably, because with

58:24 Docker, the idea is like, well, you want it, you want it isolated. So let's do this. Let me give you an

58:28 entire separate copy of Linux. Right. I know you're running it in like a, not a full VM way, and it's not

58:35 as heavy as a VM, but it's still, you're configuring a Linux computer inside of this container in the way that,

58:41 whereas this is just like, I just want Python configured, not everything.

58:45 Well, and even more like in Docker, if you want to make sure that you run Docker build twice and get

58:51 the same package versions, like you have to do that yourself. You have to come up, you have to use,

58:55 I don't know, pip compile or something like create a lock file and then install from that instead of

58:59 your original requirements. And, you know, it's a whole thing. There's lots of ways to get it wrong.

59:04 Whereas in POSY, the way I've written it currently, it's just like, there is, you know, there's one

59:09 operation, one internal function that takes a set of like, okay, these are the packages I want.

59:13 And it like renders that down into a lock file. Like, here's the exact set of packages you need,

59:18 including all the dependencies and all their versions. And then that's the thing that you hand

59:22 to the run me an environment. So like, you have to go through that step. It's just built in. And so we

59:26 can like, you know, so, and of course, you know, as we build up the, you know, CLI and stuff,

59:31 idea will be that that will be that, you know, written to disk, similar to a cargo.lock or poetry.lock

59:35 or whatever. And so you just automatically get that reproducibility, which, you know, you don't

59:41 get that automatically from Docker, right?

59:44 This is a thing that people could go get.

59:46 Yeah, on your GitHub profile, they could check this out. And they could they could try it, right?

59:51 So yeah, let's see. So yeah, so obviously, there's a lot of moving parts here. If folks want to help

59:59 with the PI BI part that PEP 711, there's lots of stuff could use help with. But it is also there's a

01:00:06 draft pap up and I have built lots of PI BI packages for lots of different versions of Python for Mac,

01:00:12 Windows Linux, they're up on a CDN. So you know, if you use the same API as PI PI, so you could pip install

01:00:19 from there if you had a pip that did it. So it's it is stuff you can try out right now, experiment with

01:00:24 at least. And then yeah, as for the posy part, again, like I said, it's still it's mostly the

01:00:29 back end stuff, but it is a pretty complete implementation of like all the packaging stuff.

01:00:33 It had it can actually do that demo. I was just saying of like, I need these three packages in this

01:00:38 version of Python, and it can do the dependency resolution for a named specific operating system,

01:00:44 which may not be the one you're running on, and then generate that environment and actually invoke

01:00:48 it. Well, you can only invoke it if it's for the operating system you're running on, of course.

01:00:52 But you know, it can do all that stuff. There isn't really a UI in front of it yet.

01:00:56 But so it's not like something I'm suggesting you go start, you know, rolling out to your company.

01:01:01 Everyone adopt it. Yeah.

01:01:02 But if this is like an exciting project for you, then you could check it out, see where it's at,

01:01:09 join in, whatever. There's definitely tons and tons of stuff to do. But it's, you know,

01:01:16 there's a good solid start. And I think it's at this point, I'm pretty confident like everything

01:01:20 could work, right? You know, kind of the proving it out part is pretty much there, done.

01:01:26 So question from Marwan in the audience. Hypothetically speaking, does that posy.lock

01:01:31 work as is on different platforms?

01:01:33 Right? Yeah. So cross platform support is a huge issue with locking. I don't know if anyone's ever

01:01:41 tried to do this with like pip compile, like it just doesn't work. If you have anything complicated,

01:01:46 like multiple Python versions, like it just doesn't work. You know, good. They tried hard. But yeah.

01:01:51 So what I'm doing right now in posy is I've tried to kind of keep it simple. I just say like,

01:02:00 tell me which platforms you care about, you know, like, you know, you know, recent dish Linux and

01:02:07 Windows 64 and macOS arm and Intel or something, you'll give me like a list. And then it will,

01:02:15 it can go through just like loop through that. And for each one, find the right PI BI, look up the

01:02:19 metadata to figure out which kinds of packages are appropriate to install there and generate a lock

01:02:25 file for each of those. And then, you know, you can somehow like merge the common parts and write them

01:02:29 to a file. So the individual things that like, you know, resolve this set of versions or set of package

01:02:37 requests into a exact set of versions that only runs at for one specific platform at a time. But you know,

01:02:45 you can run it multiple times. There are, it might be possible to do something smarter. So I know like

01:02:52 poetry has some algorithm that I don't really understand very well, where they try to sort of

01:02:57 simultaneously resolve all the platforms into their lock file. And then the way the lock file works is

01:03:02 then you actually you've, it's like only like mostly resolved. And then when you actually go to install,

01:03:08 it does like that last step to try to narrow it down to the exact platform you're on. And I just,

01:03:14 I don't quite, no one's been able to explain to me at how exactly that algorithm works or even like,

01:03:19 I'm not a hundred percent sure it's even like if it's fully correct or if it's heuristics based or

01:03:24 what. So I don't know, but like, you know, I, we can change the, there's lots of options, right? You

01:03:29 know, we can change the code if there's a better way to do it. Just that's where I'm at so far.

01:03:34 Yeah.

01:03:34 But so there's something that people can play with, but it's early days and you wouldn't mind having

01:03:38 help if people wanted to jump in.

01:03:40 No, for sure. If you want, if you've been looking for an excuse to learn Rust,

01:03:43 if you want to, yeah, play around with, you know, cool. I mean, there's interesting problems

01:03:50 in terms of things like how you efficiently resolve, you know, do package resolution.

01:03:55 It's like this whole messy, like logic programming problem. it's, you know, MP complete.

01:04:03 there's just like interesting system engineering problems of like, okay, if we're going to really

01:04:06 make this a really nice to use, like you want it, like, how do you unpack 20 wheels as fast as possible?

01:04:12 You get to use threads and concurrency and all kinds of stuff. yeah, there's lots of,

01:04:16 lots of cool technical bits too. and of course, just making something that's like a joy to use,

01:04:21 fits nicely in your hand. Lots of fun user interface problems. Yeah. I'm excited about it.

01:04:26 If you can probably tell, like, I just, I just love kinds of problems. So yeah, absolutely. I,

01:04:31 so this announcement was on discuss.python.org and I thought, okay, well, it says there's 72 responses.

01:04:39 Let me flip down and see how this landed with people. Right. Okay. Yeah. Okay.

01:04:43 It was, you know, a wide variety of responses. Yeah. Yeah. Well, but I mean, I would say that at

01:04:48 least the top batch, the first bunch of people, Paul Moore deeply involved with pip jumps in and says,

01:04:53 this is beyond awesome. I had realized you were working actively on this. I'll take a look.

01:04:58 I'd love to help out too. Talks about rest a little. Frederick says, really nice to see this.

01:05:03 This is a great direction. Giannis says, well, certainly blew my mind. Count me in on how we could

01:05:09 explore how this might work for conda and so on. And just, I thought it was really, really quite positive.

01:05:16 of how many, you know, next person. This checks many of the boxes of what I have in mind.

01:05:19 So I, it seems like it's landing well with the community. I hope that, I hope that it continues

01:05:25 to make good progress. Yeah. I think the biggest thing is like, there were definitely some folks

01:05:29 going like, okay, but why are you writing everything in rust? Like, especially it's like, you know,

01:05:35 we've spent a lot of effort, not just making standards for Python packaging, but also like implementing

01:05:40 those. So like you can pip install packaging and like right now, and that is a library that

01:05:45 could do things like unpack wheels and access the pipe. I, or I forget exactly which set,

01:05:51 but like a lot of the tricky stuff, you know, parse Python metadata, formats and, you know,

01:05:56 just all these different tricky things. And it's like, why are you re-implementing this? And also,

01:06:01 does it like send the wrong message that like, you know, when we wanted to do something complicated,

01:06:06 we thought pipe, I will, or sorry, Python wasn't good enough. We needed to switch to rust. And I get

01:06:11 where they're coming from. but well, I mean, there's a few things. So one is just that

01:06:17 I thought writing in rust would be fun. You know, I'm not telling you, you can't use Python for

01:06:21 anything. Sure. I mean, we'll, let's take a step back and say, yeah, how would, how would you propose

01:06:25 writing that in Python? Well, so there is no Python, right? That's no, no, it's possible. Right. So

01:06:30 like, so condos written Python, right. But, and the way, so it makes the distribution a little

01:06:34 complicated because like when you, you get your conda.sh or mini conda, you know, like the installer,

01:06:40 it has a Python package inside it, which it unpacks. So it uses it to run conda to install

01:06:45 like another, another Python or whatever it is that you want to install with conda.

01:06:49 but it sort of has one built in and you could do the same thing for something like posy.

01:06:53 That's something like pie installer to build an executable.

01:06:57 Exactly. Yeah. But then for my installer effectively, right. Like, like, just, you know, kind of recursively do that. it's totally, you know, it's totally

01:07:07 something one could do. the, so yeah, but the main reasons, I'm not to go that way is one is,

01:07:14 like I said, like, it's just was more interesting to me. It's one thing. There's also like, you know,

01:07:18 every, language has trade-offs, right. And the exact set of things you want from a package

01:07:25 installer are kind of right in Rust sweet spot and not pythons. So there's sort of

01:07:30 four things that are really important for something, a tool like posy. So it's, there's the initial

01:07:35 install there's, the, how quickly it starts up because like, this is in between you and invoking

01:07:43 Python or whatever it is you actually want to do. there's how quickly you can resolve packages.

01:07:49 and how quickly can it unpack packages? Like, those are the things that, you know, you care about.

01:07:54 Like those are the big load bearing pieces. And those are like kind of four of Python's weakest spots,

01:07:59 honestly. Right. Like, so like we just talked about the deployment part. you can make it work,

01:08:03 but you know, it's not as straightforward as some things. There'd be more possible things,

01:08:07 moving parts, things that could go wrong. It's not the strongest argument, but it is there.

01:08:12 for startup speed, just notoriously when a Python's weak spots, because if it has to like,

01:08:16 do all those imports from scratch every time, something, think tools like Mercurial have struggled

01:08:20 with a lot. you know, lots of Python applications doesn't matter, but you know,

01:08:24 for this particular one, that, that would be a challenge. and then, resolving is big,

01:08:30 heavy, like that MP complete, like just really gnarly burning as many CPUs as you can on complicated,

01:08:36 logical, operations. Again, not Python's strongest point. It's not something you could

01:08:41 like use NumPy for, and it's not like, you know, I'll down or anything. Yeah.

01:08:45 Like you can do something like Cython potentially, using no Gil. Yeah.

01:08:50 But then you're basically writing it in C at that point. Like you're, you're pretty far from core Python.

01:08:56 Yeah. Yeah. And then, and then finally they just, you know, unpacking files. This is totally IO bound.

01:09:01 And so simple that it's actually a big advantage to be able to just like, like, like IO is so fast

01:09:07 these days with like SSDs and everything and NVMe drives that like almost any overhead in the unpacking

01:09:14 path actually is pretty substantial as a relative proportion. So like you, you add like one Python

01:09:20 operation per, you know, you know, a hundred kilobytes written. And that might suddenly be like a 2x

01:09:26 slowdown just because everything else is so fast that even that small amount of Python overhead,

01:09:31 could be large. And, you know, in a tool like this, like people are really sensitive. Like

01:09:34 they really care if it takes 10 seconds versus one second to unpack those, the environment.

01:09:40 Like that's just a huge difference in usability. So I think it's just, it's kind of really an,

01:09:45 like, it just happens to be exact combination things that makes Rust pretty compelling, but you

01:09:50 could do it the other way too. You know, I'm not making a, it isn't meant as a political point.

01:09:55 Yeah. Okay. Got it. All right. Yeah. Python is written in C.

01:09:59 Yeah, sure.

01:10:00 So it's a PYPY, but I mean, the, the, the core, you know, the core bit of it is written in C,

01:10:07 right? No, but yeah. PYPY is written in Python, right? Hence the name.

01:10:10 Yeah, that's true. Yeah. Yeah. C PYPY written in C. It says it right there.

01:10:15 Yeah, exactly. Okay. Interesting. I think we're probably out of time to, to dive much further in

01:10:24 this, but sure. Yeah. I think we covered a lot. I think we covered a lot. I guess.

01:10:28 Anything else you want to give me your thoughts on the future. Like, what do you think? Does,

01:10:32 is the PEP gaining traction, right? This is in draft mode. I don't know how much I emphasized at the

01:10:36 beginning, but it is not an accepted. Ah, yes. That is important to be clear on that. Yeah.

01:10:40 Yeah. It's just, I wrote it. I posted it for feedback. That's as far as it's gone. It, you know,

01:10:45 there's no, there's no commitment on anyone's part to like that. This is what's actually going to

01:10:50 happen. That said, I'm pretty optimistic. Like, like I said, I got a little bit of pushback

01:10:56 on the posy part because of like the rust and whatever, but, I don't think I've gotten,

01:11:02 I can't think of really like any pushback on the PEP 7.11, the actual PYBI part, except people are

01:11:08 like, well, but why aren't you using conda or something? Which, you know, fair question, but you know,

01:11:11 there's an answer and I don't think anyone like, like it's not something that like the people that

01:11:17 you need use agreement, you need to get this accepted or like, you know, the PYPI maintainers

01:11:21 and Python packaging maintainers, and they are totally okay with like conda is not the solution

01:11:25 to everything, obviously. Right. Sure. Well, how complicated would it be to have

01:11:29 bit conda into this particular use case? Right. Yeah. I mean, there's definitely room to collaborate

01:11:34 better there. and I would love to see that in the future, but, but yeah, but my sense is that

01:11:39 there's just really hasn't been a lot of like, people just seem pretty much like, yeah, this is

01:11:43 cool. I guess actually the biggest thing is that there's been, some feedback from folks like the

01:11:49 PY oxidizer folks saying like, Hey, we would like a bit more metadata so we can fully, you know,

01:11:54 to solve some of our other things we want to do. Like we want to be able to cross compile for a given

01:11:59 Python and we need to know a bit more about the target Python in order to do that. so that's just a

01:12:03 very technical, it's like, yeah, okay. More stuff we should add and tweak.

01:12:06 It's not against the idea of evolving it. Yeah. I think the basic idea, generally people seem to be

01:12:12 on board. I'm not going to like, you know, make a commitment to what like Python.org and PYPI

01:12:17 and all they're actually going to do, but my, I don't, I'm a pretty hopeful. I think that, you know,

01:12:23 they're definitely some of the folks involved in like building the Python.org downloads right now

01:12:27 are like, Oh yeah, I'd build one of these if that was standard. Sure. so it is, it isn't all signed off

01:12:32 on and you know, full, but there seems to be a pretty reasonable consensus that like,

01:12:36 this is a good direction that we're interested in moving in. So, well, it sure caught my attention

01:12:40 when I saw it. So, I'm excited to see. Yeah. Well, thanks for having me on to, yeah. Talk about it.

01:12:45 Yeah. You bet. No, let me just ask you real quick, the final two questions since it has been five years

01:12:50 since I asked them of you. Python code. Do you work on this? what editor are you using these days?

01:12:56 I am using Emacs same as I've been since I was 13. So that's not a political position. That's just, I, you know,

01:13:04 I'm stuck. Like all your commands to come in chords, right? Okay. You know, like that's it. Yeah. Excellent.

01:13:11 And then notable PYPI package. just some random PYPI package. Yeah. Just something you like ran across.

01:13:17 Gosh, that's awesome. People should know about this. It could be very popular or not popular at all.

01:13:21 Oh man. Shoot. I did not prepare for this. Should've. I mean, I don't like, there's some obvious,

01:13:27 like obviously I like, you know, trio has been thinking about a lot, but, that's not an

01:13:31 interesting answer for this. you mentioned rough earlier. I think you mentioned rough,

01:13:37 but rough is pretty cool. Maybe I didn't mention rough. Yeah. Okay. rough is very awesome.

01:13:41 if anyone doesn't know rough is sort of, you know, flake eight, and such re-implemented in rust. So it's like

01:13:47 a hundred times faster. Like you just like instantaneously lends all your code, which is

01:13:51 very sweet. There's a selling point for rust integrated with Python, right? Like

01:13:55 another use case that looks pretty neat. Yeah. And I, you know, sort of, as I'm digging into it,

01:14:00 I'm really impressed at how they, those two, how well they fit together. People put a lot of work

01:14:05 into like making that really smooth and having them collaborate. Well, actually it's something I've just

01:14:09 been working on at work is, we've been having trouble with, so, in an async library like

01:14:16 trio, the, you have lots of tasks running concurrently, but they only like the scheduler

01:14:22 only gets to switch from one task to another when one task explicitly lets go, like says, okay, I can

01:14:27 stop here. We're using an await statement. so it's possible to write code where you accidentally

01:14:32 like don't do that for a long time. And that task will just like hog, all the runtime and block

01:14:37 other tasks from running. And it'd be nice. It's hard to kind of tell when that's happening. and

01:14:41 a similar thing could happen with the GIL. so if you have like an extension library, like, you know,

01:14:46 PyTorch or something, and they forget to drop the GIL before doing some big heavyweight operation,

01:14:50 then it could just block any other threads from running. that's really awkward. We've been

01:14:54 having trouble with that. And, but well, you know, like PySpy, there's another really cool package.

01:14:59 If anyone's seen it, is a rust profiler for Python that can just sit outside your process and,

01:15:04 can, like give you, you know, tell you what it's doing. and, but also it being in rust

01:15:10 and it's up on, you know, crates.io, I could just write a little program that imports, PySpy

01:15:16 is use it as a library, but, and tweak it. So that instead of looking for like, where's code

01:15:21 spending time, it detects, okay, is something hogging, you know, the GIL or the run loop and,

01:15:27 and, and give me the trace back, show me which code is doing that. and it's again, really neat to be

01:15:33 able to, you know, get that really deep insight into this Python stuff that, you know, we're still,

01:15:37 you know, we were using it. It's still Python, but, the rust really is a great flavor that goes with it.

01:15:43 Cool. PySpy. All right. People can check that out. That's Sampling Profiler for Python programs. Yes. Yeah. PySpy is really cool.

01:15:51 Indeed. All right. Well, okay, cool. Thanks for being here. if people are interested in the pep,

01:15:56 what should they do? I mean, I guess the, the post on discuss.uh, python.org,

01:16:03 is the best place for like feedback. It's also, where I posted about, Posey. So if you want to

01:16:09 like see the discussion or join in, that's a good place. If you want to like, you know, help then yeah.

01:16:15 GitHub.com/njsmith/posy is the repository, you know, jump in, send PRs, file issues, whatever.

01:16:22 or just, you know, send me a, you know, I don't know. Yeah. What's the toot at me, I guess,

01:16:27 the Mastodon version. I'm not really on Twitter these days, but yeah. njs@mastodon.social.

01:16:34 and I'll see it. Cool. All right. Well, Nathaniel, thanks for being here. Thanks for this

01:16:38 pep. It looks interesting. Yeah. Thanks. It's great. Great being here. Yeah, you bet.

01:16:43 That this has been another episode of talk Python to me. Thank you to our sponsors. Be sure to check

01:16:50 out what they're offering. It really helps support the show. Take some stress out of your life. Get

01:16:55 notified immediately about errors and performance issues in your web or mobile applications with

01:17:00 Sentry. Just visit talkpython.fm/sentry and get started for free. And be sure to use the promo code,

01:17:07 talk Python, all one word. Want to level up your Python? We have one of the largest catalogs of

01:17:12 Python video courses over at talk Python. Our content ranges from true beginners to deeply advanced topics

01:17:18 like memory and async. And best of all, there's not a subscription in sight. Check it out for yourself

01:17:23 at training.talkpython.fm. Be sure to subscribe to the show, open your favorite podcast app and search

01:17:29 for Python. We should be right at the top. You can also find the iTunes feed at /itunes, the Google Play

01:17:35 feed at /play and the direct RSS feed at /rss on talkpython.fm. We're live streaming most of our

01:17:43 recordings these days. If you want to be part of the show and have your comments featured on the air,

01:17:47 be sure to subscribe to our YouTube channel at talkpython.fm/youtube. This is your host,

01:17:53 Michael Kennedy. Thanks so much for listening. I really appreciate it. Now get out there and write

01:17:57 some Python code.

01:17:59 it's just a couple of tips on the show. I'll see you next time. I'll see you next time. I'll see you next time.