Re 1:27:00ish The cultural norm in OCaml is no type annotations in `.ml` files, but typically people write `.mli` interface files, and there all public signatures are annotated
45:46 so, slowly programing languages are converging into Lisp. Imagine decades of career experience just to converge back in 1959. Just another case like Uncle Bob.
Chris seems like a smart guy so his ignorance regarding Rust is a bit surprising. Rust panics on integer overflows in debug mode because it gives you at least some chance of detecting the error early. In release mode Rust uses wrapping just like Mojo. So it's really the best of both worlds. If you always want wrapping behavior you should use the special methods for that. And usage of unsafe is not frowned upon in the Rust community. Of course it's ok to use it in some cases (like the example they bring up of interfacing C libraries), but you very seldom need to use it for performance reasons (array bounds checking is typically avoided by using iterators).
"ignorance regarding Rust" why would he care about a seemingly useless language in the context of his professional career? I'm willing to bet his priority list doesn't include Rust. I don't take Chris for a language fanatic / fanboy so this is not surprising.
I really wished Chris worked on something more open that takes into account all the mistakes and techniques we’ve learned from languages such as Python, C++, Rust, Nim, Zig, Odin (even Jai) etc including Swift (as Chris himself said in ThePrimeagean stream, That Swift became an unnecessarily complex language) and paradigms such as OOP, FP etc and made something wholly fresh and new that is GREAT. But who has the money to pay except all of those VCs that are hyped currently for “AI”? 😢
Modular is building a managed cloud offering for AI deployment. Their license disallows building specifically just that software without written permission. Can't say I can fault them, there are just a lot of AI cloud popping up and each seeking advantage, making something you intend to turn into a business without a good licence could decimate your entire company before it can stand on its own.
Respectfully, that's pretty disingenuous. Chris spent his whole life building some really cool open source projects like LLVM/Clang and employed at companies building up their IP, and now he wants to build something for himself and God forbid be able to make a living on it and be able to employ high-quality people to work on it so they too can earn a living, and that's a "cash grab"?
Any serious programming language that aims at the future must limit what is built into the compiler and prioritize being meta-first. We have an abundance of programming languages, but for mostly poor reasons. Managed languages, in my opinion, have no future. They lost their edge when Rust emerged. While Rust still has some rough edges, it is the easiest language to write performant, safe, and self-contained programs. Languages like Java, C#, Kotlin, Python, and Ruby, among others, fall more into the category of DSLs. As for Swift, sorry Chris, but it feels like a mishmash that tries to combine too many elements that don’t quite fit together. Its programming model seems overly chaotic. By the way, OOP was and still is a terrible idea. I hope this ideology meets a swift end!
@@familyshare3724 definitely legacy, but what I meant is that they cannot be considered general-purpose if you are serious about programming, because their programming model is very poor and limited.
![Mojo 🔥 - Is it Python's faster cousin or just hype? [PyCon DE & PyData Berlin 2024]](http://i.ytimg.com/vi/A67P4-A4yLk/mqdefault.jpg)
I highly appreciate this technical stuff with detailed questions! Great podcast, thank you :)
this is highly technical but approachable
I love these podcasts, they always motivate me to continue working on my language, thank you for that!
what is it? 😊
igorlang
Re 1:27:00ish
The cultural norm in OCaml is no type annotations in `.ml` files, but typically people write `.mli` interface files, and there all public signatures are annotated
Really good discussion. MAX/Mojo sound super awesome.
Awesome!!!
I suppose then signature annotation are the same in Python and Mojo (not list[int]) but def myF(i: List[Int]) -> Int:
Raku also has excellent Unicode support with native support for grapheme clusters
45:46 so, slowly programing languages are converging into Lisp. Imagine decades of career experience just to converge back in 1959. Just another case like Uncle Bob.
Chris seems like a smart guy so his ignorance regarding Rust is a bit surprising. Rust panics on integer overflows in debug mode because it gives you at least some chance of detecting the error early. In release mode Rust uses wrapping just like Mojo. So it's really the best of both worlds. If you always want wrapping behavior you should use the special methods for that.
And usage of unsafe is not frowned upon in the Rust community. Of course it's ok to use it in some cases (like the example they bring up of interfacing C libraries), but you very seldom need to use it for performance reasons (array bounds checking is typically avoided by using iterators).
"ignorance regarding Rust" why would he care about a seemingly useless language in the context of his professional career? I'm willing to bet his priority list doesn't include Rust. I don't take Chris for a language fanatic / fanboy so this is not surprising.
@@GuiseppeZanotta totally agree, People like this really do make rust feel like a cult
even Chris doesn't use LLVM 😂
Not touching Mojo until its completely open source 💯
This. Not touching Mojo until they get rid of the competitive clause (for Mojo. Idc the Max platform).
we'll try to battle forward despite this great loss i really hope we can get by without you and yours. 😂
my favorite part about learning Swift is it made Rust no biggie 😅if you dont think you can, you're wrong.
its still isn't a superset, calling it a superset is just a buzz word, C & C++ are two programming languages with a big intersection.
Whats with mojos licensing? It makes it seem more like an AI cash grab than a trustable language
ruclips.net/video/tYujSI_aoqk/видео.htmlsi=1c43b3mPcFMHqukJ
I really wished Chris worked on something more open that takes into account all the mistakes and techniques we’ve learned from languages such as Python, C++, Rust, Nim, Zig, Odin (even Jai) etc including Swift (as Chris himself said in ThePrimeagean stream, That Swift became an unnecessarily complex language) and paradigms such as OOP, FP etc and made something wholly fresh and new that is GREAT. But who has the money to pay except all of those VCs that are hyped currently for “AI”? 😢
Modular is building a managed cloud offering for AI deployment. Their license disallows building specifically just that software without written permission. Can't say I can fault them, there are just a lot of AI cloud popping up and each seeking advantage, making something you intend to turn into a business without a good licence could decimate your entire company before it can stand on its own.
Respectfully, that's pretty disingenuous. Chris spent his whole life building some really cool open source projects like LLVM/Clang and employed at companies building up their IP, and now he wants to build something for himself and God forbid be able to make a living on it and be able to employ high-quality people to work on it so they too can earn a living, and that's a "cash grab"?
A more charitable take would be that Chris is trying to create a sustainable funding source to support work on Mojo
Any serious programming language that aims at the future must limit what is built into the compiler and prioritize being meta-first. We have an abundance of programming languages, but for mostly poor reasons.
Managed languages, in my opinion, have no future. They lost their edge when Rust emerged. While Rust still has some rough edges, it is the easiest language to write performant, safe, and self-contained programs. Languages like Java, C#, Kotlin, Python, and Ruby, among others, fall more into the category of DSLs.
As for Swift, sorry Chris, but it feels like a mishmash that tries to combine too many elements that don’t quite fit together. Its programming model seems overly chaotic.
By the way, OOP was and still is a terrible idea. I hope this ideology meets a swift end!
domain specific or legacy specific?
@@familyshare3724 definitely legacy, but what I meant is that they cannot be considered general-purpose if you are serious about programming, because their programming model is very poor and limited.