Here are the timestamps. Please check out our sponsors to support this podcast. 0:00 - Introduction & sponsor mentions: - iHerb: lexfridman.com/iherb and use code LEX to get 22% off your order - Numerai: numer.ai/lex - InsideTracker: insidetracker.com/lex to get 20% off 2:20 - Mojo programming language 12:37 - Code indentation 21:04 - The power of autotuning 30:54 - Typed programming languages 47:38 - Immutability 59:56 - Distributed deployment 1:34:23 - Mojo vs CPython 1:50:12 - Guido van Rossum 1:57:13 - Mojo vs PyTorch vs TensorFlow 2:00:37 - Swift programming language 2:06:09 - Julia programming language 2:11:14 - Switching programming languages 2:20:40 - Mojo playground 2:25:30 - Jeremy Howard 2:36:16 - Function overloading 2:44:41 - Error vs Exception 2:52:21 - Mojo roadmap 3:05:23 - Building a company 3:17:09 - ChatGPT 3:23:32 - Danger of AI 3:27:27 - Future of programming 3:30:43 - Advice for young people
The Stone Henge and Pyramids, etc all were easy to move during these time windows of lost Human history because of the low levels of gravity due to the Earth's Axis tilt was different and the Moon also played a key role... Question: How would humans today, go about moving massive stones on the Moon today? These large stone structures were carved and relocated over miles from their origins; they were moved with large animals pulling ropes, dragging them like large foam blocks, leaving little trace. Left the future gens boggled... I drew out diagram. It's the only thing that logically fits.
ChatGPT will conclude this in near future - facts of our lost human history in regards to the low levels of gravity - how we moved these massive large stone blocks and statues etc...
If Jeremy Howard is saying in his fast ai blog "Mojo may be the biggest programming language advance in decades" then it's a very big deal, and I'm paying attention. Thanks Lex for another fantastic interview. Thanks Chris for the clear explanations and of course, all the work you put into Mojo.
@@OfTheVoid Also, the reason the folks during low gravity era used large heavy stone blocks is because, they would stack up - weigh each other down - everything down - otherwise, they could have just used small blocks - bricks and achieve these structures and stutues - everything moved at ease... done in the fraction. The core of the Earth has a ball within a ball(rotating opt direction,) the core of core rotation is decressing and will pause and then rotate the other direction... this could also a art of the low gravity that occurred 30,000 + years ago.
Absolutely. This channel is a mad superstar VIP party for nerds. Every time I am modeling something in Blender, or even doing chores or anything where language / listening doesn't conflict with the task itself, I know I will be able to find something that is continuously engaging and interesting on this channel. Oh and, thank you for essentially painting my living room. I was listening to another episode as I did that, which resulted in my brain delegating the utterly boring task to the spinal cord entirely, which lead to me being basically unaware of having painted my room (I only remember the discussion, and the fact I had sore muscles the next day).
This guy just walks around fixing programming languages and compilers, From Clang to Swift and now Mojo...God knows what he'll be doing next... An OS probably. We are lucky to have him in humanity.
I don't see him in building an OS. While he is exceptionally diverse he always improved programming languages for hardware or the other way around. But i wouldn't surprise me if he makes an OS specific language that get's adopted by the Linux Kernel maintainers because it's just that good.
Let's help make it a more perfect system. There are always areas a slick eye can pick up on that no other can, and if you're not obnoxious and short-sighted certainly the team can weigh your thoughts in with that of the collective mind.
as a professional programmer for the last decade, listening to Chris is mental. He is so amazing. This conversation was one of the best things I've ever listened to
Love Chris. Such a good guest every time. He is the epitome of a guest that is clear and concise in his delivery despite him having a wealth of knowledge.
@@Hexanitrobenzene I usually default to that stuff when falling asleep actually 😁 I just happened to be listening to it, and i know she falls asleep sometimes even when my wife and I just talk to each other with her in the bed. Maybe she found the podcast interesting in some way :)
well done dad...my daughter is 6 also and we love listening to Lex on our daily commute together. Different topic but I just shake my head seeing some girls in her class wearing make up and miniskirts
Chris Lattner’s CV is so legendary. I think this is the longest intro Lex has given for a guest on the show, and I think he realized it in the middle of listing Chris’ accomplishments 😂
When I discovered Python about 15 years ago, I was so jazzed - and have been using it for countless projects, commercial and otherwise. Then I discovered Swift and SwiftUI over a year ago, and, for completely different use cases, have really been impressed. Now, here comes Lex interviewing Chris Lattner once again, about his latest foray into improving Python in so many ways. Guido tried to do some of these things when he was at Google, but Chris might just be able to pull this off. This is groundbreaking! Thank you Lex!
@@aoeu256 Damn, I haven't seen any other human endeavour where there is so much reinvention of the wheel as in programming... Julia was invented for this exact purpose, to be concise as Python and fast as C. Lisp is like a 3D printer for Domain Specific Languages., but damn... It's for nerds only :)
@@Hexanitrobenzene its a chance to fix the countless mistakes julia made: one based indexing, column major, atrocious import system which defaults to "from module import *" the (resulting?) lack of good static analysis, the stupid idea of whitespace semantics and finally the complete lack of any formal interface (which means zero guarantees than anything works as intended) I hope that Julia ends up in the history books a badly written mock up for mojo
When I moved to Germany about 20 years ago, before I understood the formality of address, in rules of German language, I thought it was beyond hilarious that people with advanced degrees were addressed with both the gender and degree when they were referred to. For instance “Mr. Doctor” then Lastname. I came to understand it eventually and it’s intent to show respect. So for me, from now on, you are “Mr. Doctor Lex.” Your interviews are simply outstanding - I’ve learned an incredible amount - my 65 year old brain just about can’t wait for the next episodes. Thank you!
Having an engineer as a manager really helps since he is able to understand the nature of the problem he is trying to solve. It's like old school boeing and bell labs. I found this podcast by complete accident and ended up watching the whole thing - so much depth and great content, even for someone who isn't a professional in this field.
If Jeremy Howard is saying in his fast ai blog "Mojo may be the biggest programming language advance in decades" then it's a very big deal, and I'm paying attention. Thanks Lex for another fantastic interview. Thanks Chris for the clear explanations and of course, all the work you put into Mojo.
Frankly, for all his engineering acumen, which is clearly amazing and worthy of praise, his review and deep consideration of the psychological & sociological impact of AI & AGI was incredibly shallow & almost restricted by a 'free mrkt.' worldview confined by engineering - those are his tools so I appreciate that at one hand but he has clear blindspots that are problematic when considering the ramifications of AI and AGI...not from the engineering paradigm but from their impact and on that there are far more deep thinkers. An astounding engineer though - truly one-of-a-kind.
To be fair he is building rock solid building blocks for intense computation. His understanding of AI and it's impact is surely more profound than most peoples, but this has barely anything to do with his profession of building good programming languages. It's like asking a civil engineer for bridges what impact car emissions have, because he enables them to go fast in a large scale.
Dang it! I was waiting for a new round! Chris is really an inspiring person. How to tackle the big issues in software programming…on his words really encouraged me to follow his projects along his professional life. Thank you very much, Lex for this new round!
As an newbie to this field, this is so inspirational yet so intimidating. So much breadth and depth in the field of computer science. One lifetime isn’t long enough for all the cool possibilities. What a time…
I really want to listen to this tonight! But, I have a four hour drive tomorrow, and I'm thinking this would make the drive much more fun. Added to watch later.
I always love the programming related ones and im glad I'm able to geek out over the convo the same way they are 😂 obviously nowhere near as good as these 2 but definitely as passionate. Thanks lex 👍
as a compiler writer, chris has always been my fave guest. all the way since the ai pod days. thanks for another amazing pod you two!! here’s to mojo 🔥🍻
I like Chris, but the argument is at least partially disingenuous. For example, you still need auto-formatting in Python for standardization on projects, since indentation amount and type isn't forced by the language. And he didn't mention anything about the multitude of linter/formatter options that really make the difference in readability and reliability - curlies or not - that motivate using them on projects. Sadly, I see his willingness to sell religion as objectivity as undermining his opinions on the areas I'm really watching this to hear about. It's clear he enjoys poking people, so the overstatement is certainly intentional. Regardless, he usually has interesting stuff to say among all the BS, so I'm enjoying the conversation still overall.
@@rob3c Every (good) language has formatters and linters. That's not his point. If you are going to indent anyway cause its easier to ready, why add the curlies?
You're right about the tech language Lex. As a lay person, I don't understand enough in this episode but I do enjoy the excitement and inspiration from two pros. It still sounds like music to my ears. Thank you so much as always!
Can't remember when was the last time I thoroughly enjoyed a long conversation on the interwebs as much as I did this one. Thank you both, that was awesome!
enjoyed the whole thing so much. I'm so with Chris on complexity being THE enemy ... took me a long time to learn. Python was my first love and I still love it. However since I found Erlang/OTP and the BEAM runtime I've come to believe this is the strongest programming env and runtime in a surprisingly large number of domains. especially when combined with a language like Elixir which has been designed for similar goals as what Guido had in mind for Python. The fundamental message-passing concurrency, the error handling philosophy, and now even the ML capabilities with NX and higher-level libraries. Joe Armstrong's thesis made me realize how much of a secret sauce we've got in our hands. As Chris days "when everyone goes left, you sometimes have to go right". Python might be the clear winner in the AI/ML space for now .... but I'm betting it might not be forever 😉
Great talk, unfortunately he misinterprets the "zero cost" meaning in "zero cost exceptions" (and it's *not* zero cost exception *handling*), the same way many people misinterpret the term zero cost abstractions. The zero cost refers to the fact that you don't have any additional runtime cost, specifically in the happy-path, meaning that if a function doesn't throw, there's no downside of having the `throw` in the function itself from a performance perspective. But ofc depending on your definition of "cost" you'll still get larger binaries, require RTTI and so on, but that's usually not what people find relevant in 99.9% of cases. It's the exact same thing regarding zero cost abstractions, you have tons of abstractions that don't affect the runtime, but ofc it'll affect lesser things such as compile times and what not. And talking about returning values that represent errors like you do in Rust, or with `std::expected` since C++23, that's not relatable to exceptions. Exceptions and result types cover different areas of handling errors that aren't necessarily interchangeable, hence having the support for both is optimal.
he make a good point about the file extension and looking at them, took me a week or two to remember the order of .ipynb properly. And I've very stoked to see more from mojo.
I loved any minute of it. If someone can solve this complex AI problem, then it is most probable Chris. Playing around with Mojo from its infant stages feels like being part of the history. Thank you, Lex! Thank you, Chris!
I love how excited Lex is for the conversation. He's a programming nerd like me. This stuff is fascinating even though I don't fully understand everything...yet.
The main selling point for mojo is that you can leverage MLIR to write custom operators instead of relying on a runtime written in C++. Like for tensor operations pytorch uses Aten as the runtime which is written in C++. Mojo is like pytorch GLOW or tensroflows JAX. I don't understand the need for mojo when there are better solutions.
@@solitary_crow I think they are trying to be TypeScript for AI. I have the impression that Chris saw a bussiness opportunity for LLVM like stacks in proliferation of hardware in AI and decided to create such a stack with modular. As Python is the language of AI, they chose Python as an interface to their stack in order to attract users.
Having chris lattner on your podcast for the second time make me smile happy. James Gosling also genius person have so many wise and knowledgable experiences in the IT field. could you invite him for the second round lex?
packaging in python with things like poetry is pretty straight forward and clean, rarely problems with c or c++ dependencies (this can happen on certain platforms like windows where things might not be well tested, or new hardware like apple silicone where the compiling process might not be optimal or correct)
Phenomenal video, fell in love with hardware as a child and became a software guy as an adult, seeing it all morph into these heterogeneous systems is magical
Can someone provide any resources to become hardware literate? As a data scientist, I'm intrigued by the intersection of software/AI and hardware (which is mainly what Chris Lattner talks about), but I find it quite challenging to find sources that can help me become sufficiently literate without becoming an expert (I haven't found any course from Chris Lattner or similar, it would be nice if there is one haha)
Very interesting video. Since Chris mentioned Zig, I see a lot of parallels in the way they are operating. Zig seems to be following that Swift model to bring the C community along, while Mojo looks to be using the release early and get feedback open model Zig is using. I wonder if they have considered using Zig to solve the C packaging and compiling portion for python to help facilitate the transition? After realizing this problem, Zig seems like a natural fit for Python and Mojo for this.
Amazing. Ive been comissioning computers on site for years now and i have been watching operating systems, software, everything become more bloated, obnoxious and convoluted to use. Faster computers do indeed make for lazy programmers... Chris is the first programmer ive seen who 'gets' the horrendous inefficiencies that we have learned to put up with with our hyper fast computers - to hear he has made a language thats 35,000* faster than everyone else... It makes me think about american cars in the 60's - fuel was so cheap it was easier to just pout a bigger engine in if you wanted to go fast. Nowadays a 1L motorbike engine can make the same power as the 7.2L v8's of old because of the constraints, fuel pricing etc. We would have come up with this language ages ago if we didnt have moores law. Keep it up Chris i was just about to jump into learning python, looks like mojo now!
Please ask every guest you have for advice for young people. It is the most important segment that I always look for in a video. Please ask for advice for young people from every guest.
Love how you mention list comprehension with the things you love about python. Ever since I learned how to write them I never populated lists in the same way. I use them almost to a flaw lol
Fascinating from begin() to end()! I can see that Chris would be a great person to work with. If I were looking, I would drop him a line. I hope Modular does well.
Always fascinated as to the progress of the Human condition. You inspire me more with every new podcast. If we are to advance, it will be with the optimization of our marriage with AI, and even like a marriage things will be tense but the outcome will be forever positive.
Just finished listening to this episode over multiple sittings. Amazing content packed with fascinating insights, 120% must-see for anyone who is interested in the evolution of programming languages. Thanks so much for doing this!
![Lethal Company Music Video 🎵 "Follow The Orders" [VERSION A]](http://i.ytimg.com/vi/QeooZB4Z7RA/mqdefault.jpg)
Here are the timestamps. Please check out our sponsors to support this podcast.
0:00 - Introduction & sponsor mentions:
- iHerb: lexfridman.com/iherb and use code LEX to get 22% off your order
- Numerai: numer.ai/lex
- InsideTracker: insidetracker.com/lex to get 20% off
2:20 - Mojo programming language
12:37 - Code indentation
21:04 - The power of autotuning
30:54 - Typed programming languages
47:38 - Immutability
59:56 - Distributed deployment
1:34:23 - Mojo vs CPython
1:50:12 - Guido van Rossum
1:57:13 - Mojo vs PyTorch vs TensorFlow
2:00:37 - Swift programming language
2:06:09 - Julia programming language
2:11:14 - Switching programming languages
2:20:40 - Mojo playground
2:25:30 - Jeremy Howard
2:36:16 - Function overloading
2:44:41 - Error vs Exception
2:52:21 - Mojo roadmap
3:05:23 - Building a company
3:17:09 - ChatGPT
3:23:32 - Danger of AI
3:27:27 - Future of programming
3:30:43 - Advice for young people
The Stone Henge and Pyramids, etc all were easy to move during these time windows of lost Human history because of the low levels of gravity due to the Earth's Axis tilt was different and the Moon also played a key role... Question: How would humans today, go about moving massive stones on the Moon today? These large stone structures were carved and relocated over miles from their origins; they were moved with large animals pulling ropes, dragging them like large foam blocks, leaving little trace. Left the future gens boggled... I drew out diagram. It's the only thing that logically fits.
ChatGPT will conclude this in near future - facts of our lost human history in regards to the low levels of gravity - how we moved these massive large stone blocks and statues etc...
If Jeremy Howard is saying in his fast ai blog "Mojo may be the biggest programming language advance in decades" then it's a very big deal, and I'm paying attention. Thanks Lex for another fantastic interview. Thanks Chris for the clear explanations and of course, all the work you put into Mojo.
@@OfTheVoid Also, the reason the folks during low gravity era used large heavy stone blocks is because, they would stack up - weigh each other down - everything down - otherwise, they could have just used small blocks - bricks and achieve these structures and stutues - everything moved at ease... done in the fraction. The core of the Earth has a ball within a ball(rotating opt direction,) the core of core rotation is decressing and will pause and then rotate the other direction... this could also a art of the low gravity that occurred 30,000 + years ago.
I think ChatGPT - all this AI tech will be able to figure this stuff out.
Seriously man, you've had KILLER guests recently. Learning a lot, thank you for contributing to humanity's knowledge base.
This, absolutely.
Well we need to beat the AI
Ditto. 💯
Absolutely. This channel is a mad superstar VIP party for nerds. Every time I am modeling something in Blender, or even doing chores or anything where language / listening doesn't conflict with the task itself, I know I will be able to find something that is continuously engaging and interesting on this channel.
Oh and, thank you for essentially painting my living room. I was listening to another episode as I did that, which resulted in my brain delegating the utterly boring task to the spinal cord entirely, which lead to me being basically unaware of having painted my room (I only remember the discussion, and the fact I had sore muscles the next day).
Recently? For a long time!
This guy just walks around fixing programming languages and compilers, From Clang to Swift and now Mojo...God knows what he'll be doing next... An OS probably. We are lucky to have him in humanity.
Sort of a Nietsche's Übermensch 😂
He seems pretty central to having created the world we human inhabit
I don't see him in building an OS. While he is exceptionally diverse he always improved programming languages for hardware or the other way around.
But i wouldn't surprise me if he makes an OS specific language that get's adopted by the Linux Kernel maintainers because it's just that good.
these guys with modular are on the right path, their head is straight about what's going on, and how things should be. good to see chris again!
Let's help make it a more perfect system. There are always areas a slick eye can pick up on that no other can, and if you're not obnoxious and short-sighted certainly the team can weigh your thoughts in with that of the collective mind.
as a professional programmer for the last decade, listening to Chris is mental. He is so amazing. This conversation was one of the best things I've ever listened to
Love Chris. Such a good guest every time. He is the epitome of a guest that is clear and concise in his delivery despite him having a wealth of knowledge.
For us programming nerds, this is golden.
well thats good someone got something out of this because thats gotta be the only ppl that did. Otherwise its 3+ hours of boring fucking jibberish
Hey bro, do you luv diggs?
My daughter was having nightmares and we listened to this podcast to distract her. She asked for it again tonight, she's 6 :)😊
Hm, I would have chosen something on astrophysics, debate about intelligent life in space... or would it actually reinforce the nightmares ?
@@Hexanitrobenzene I usually default to that stuff when falling asleep actually 😁 I just happened to be listening to it, and i know she falls asleep sometimes even when my wife and I just talk to each other with her in the bed. Maybe she found the podcast interesting in some way :)
@@danielhenderson7050
She probably liked Lattner's manner of speech. It gives off positive vibes :)
😆@Bebtelovimab
well done dad...my daughter is 6 also and we love listening to Lex on our daily commute together.
Different topic but I just shake my head seeing some girls in her class wearing make up and miniskirts
Chris Lattner’s CV is so legendary. I think this is the longest intro Lex has given for a guest on the show, and I think he realized it in the middle of listing Chris’ accomplishments 😂
When I discovered Python about 15 years ago, I was so jazzed - and have been using it for countless projects, commercial and otherwise. Then I discovered Swift and SwiftUI over a year ago, and, for completely different use cases, have really been impressed. Now, here comes Lex interviewing Chris Lattner once again, about his latest foray into improving Python in so many ways. Guido tried to do some of these things when he was at Google, but Chris might just be able to pull this off. This is groundbreaking! Thank you Lex!
What I’m wondering is why people forget about Julia and Lisps in general.
Lisps are too OP
@@aoeu256
Damn, I haven't seen any other human endeavour where there is so much reinvention of the wheel as in programming... Julia was invented for this exact purpose, to be concise as Python and fast as C.
Lisp is like a 3D printer for Domain Specific Languages., but damn... It's for nerds only :)
@@Hexanitrobenzene its a chance to fix the countless mistakes julia made: one based indexing, column major, atrocious import system which defaults to "from module import *" the (resulting?) lack of good static analysis, the stupid idea of whitespace semantics and finally the complete lack of any formal interface (which means zero guarantees than anything works as intended)
I hope that Julia ends up in the history books a badly written mock up for mojo
@@trulyUnAssuming
Looks like I'm out of my depth here... Julia has whitespace semantics ?
Chris is one of those guests I could listen to all day. He’s really great at effectively communicating complex topics. Glad he keeps coming back!
I’ve been craving this since the Mojo announcement. Thanks, Lex!
One of the best geek nerdy conversations in a long time. I loved it and obvioulsy so did Lex and Chris. You can see the romance blossom.
Unless your facet of nerdiness doesn't involve advanced coding :(
When I moved to Germany about 20 years ago, before I understood the formality of address, in rules of German language, I thought it was beyond hilarious that people with advanced degrees were addressed with both the gender and degree when they were referred to. For instance “Mr. Doctor” then Lastname. I came to understand it eventually and it’s intent to show respect. So for me, from now on, you are “Mr. Doctor Lex.” Your interviews are simply outstanding - I’ve learned an incredible amount - my 65 year old brain just about can’t wait for the next episodes. Thank you!
This Mr. Doctor treats mental deficiencies very well.
Having an engineer as a manager really helps since he is able to understand the nature of the problem he is trying to solve. It's like old school boeing and bell labs.
I found this podcast by complete accident and ended up watching the whole thing - so much depth and great content, even for someone who isn't a professional in this field.
The new school Boeing is the one that ChatGPT built: hallucinating, doors falling off etc etc.
Well Bosch, Boing, Porsche, and the early guys at IBM, Intel or Apple were all engineers. The manager's didn't came until the profit was there.
Oh yes, after the announce of Mojo I couldn't wait for the next interview w Chris Lattner. Can't believe it's already here!
Will websites on web assembly and mojo be faster than JavaScript hmm….
If Jeremy Howard is saying in his fast ai blog "Mojo may be the biggest programming language advance in decades" then it's a very big deal, and I'm paying attention. Thanks Lex for another fantastic interview. Thanks Chris for the clear explanations and of course, all the work you put into Mojo.
In a noisy world, Jeremy Howard is pure signal.
Frankly, for all his engineering acumen, which is clearly amazing and worthy of praise, his review and deep consideration of the psychological & sociological impact of AI & AGI was incredibly shallow & almost restricted by a 'free mrkt.' worldview confined by engineering - those are his tools so I appreciate that at one hand but he has clear blindspots that are problematic when considering the ramifications of AI and AGI...not from the engineering paradigm but from their impact and on that there are far more deep thinkers. An astounding engineer though - truly one-of-a-kind.
To be fair he is building rock solid building blocks for intense computation. His understanding of AI and it's impact is surely more profound than most peoples, but this has barely anything to do with his profession of building good programming languages.
It's like asking a civil engineer for bridges what impact car emissions have, because he enables them to go fast in a large scale.
Dang it! I was waiting for a new round! Chris is really an inspiring person. How to tackle the big issues in software programming…on his words really encouraged me to follow his projects along his professional life. Thank you very much, Lex for this new round!
Lex fridman is best podcast host there is!
As long as he does not take up political topics he is great. Too pro Putin for my taste.
@@Koipeliini1 Lex is neutral. You are just in an Eco-chamber.
@@Koipeliini1 he's not pro Putin
quite possibly!
@@Koipeliini1 pro-Putin? lol wut?
As an newbie to this field, this is so inspirational yet so intimidating. So much breadth and depth in the field of computer science. One lifetime isn’t long enough for all the cool possibilities. What a time…
I really want to listen to this tonight! But, I have a four hour drive tomorrow, and I'm thinking this would make the drive much more fun. Added to watch later.
Such a joy to be able to listen in to some of the most fantastic conversations. The speed of growth in self learning programs is inexorable.
💚♾️
What a wise, humble and lovely person. Few leaders are like that these days.
My favorite guest on this podcast😊
Oh boy, what an awesome podcast. Seems like podcasts with hardcore software and hardware guys are the best.
Positive waves, everyone.
Negative waves are required for alternating current
I always love the programming related ones and im glad I'm able to geek out over the convo the same way they are 😂 obviously nowhere near as good as these 2 but definitely as passionate. Thanks lex 👍
I think this is about to be the coolest and best podcast i watch this year
as a compiler writer, chris has always been my fave guest. all the way since the ai pod days. thanks for another amazing pod you two!! here’s to mojo 🔥🍻
Was just checking out Mojo lang , and here we have Chris
It's rare to be such a high level of expertise and enjoyable at the same time.
Love this through and through. Also love this channel, great work Lex!
Amazing that everything goes back to C and C++ to get things done. I am interested on the Mojo for sure!
or fortran
I was interested in Mojo not anymore because you can not run locally and it needs other parts to run. Not stand alone.
@@InteractiveDNA
...yet. It's v0.1 only.
I get the feeling that this is the future of fast dynamically typeable languages.
This is really exciting.
Thanks for having this guy on, Lex.
I've been writing C# for almost 20 years. This is the best argument I've ever heard against curlies. I might actually be changing my mind. Damn.
Yeah, it's just simpler to have one thing - indentation - represent grouping.
I like Chris, but the argument is at least partially disingenuous. For example, you still need auto-formatting in Python for standardization on projects, since indentation amount and type isn't forced by the language. And he didn't mention anything about the multitude of linter/formatter options that really make the difference in readability and reliability - curlies or not - that motivate using them on projects. Sadly, I see his willingness to sell religion as objectivity as undermining his opinions on the areas I'm really watching this to hear about. It's clear he enjoys poking people, so the overstatement is certainly intentional. Regardless, he usually has interesting stuff to say among all the BS, so I'm enjoying the conversation still overall.
@@rob3c Every (good) language has formatters and linters. That's not his point. If you are going to indent anyway cause its easier to ready, why add the curlies?
@@haxi52 I understand his point just fine, thanks
You're right about the tech language Lex. As a lay person, I don't understand enough in this episode but I do enjoy the excitement and inspiration from two pros. It still sounds like music to my ears. Thank you so much as always!
This is freaking awesome. What an amazing time to be alive or a simulation or an alien.
Can't remember when was the last time I thoroughly enjoyed a long conversation on the interwebs as much as I did this one.
Thank you both, that was awesome!
Can’t wait for Mojo to run natively on Mac.
I love hearing lex talk about his passion python. You can see the joy in his eyes. Keep it up lex!
You are killing it Lex. Another interview, that I had to listen to very intently.
I am loving this.
Chris is a legend 🔥
This came at the right time. I just re-watched the 1st and 2nd episode
Nice conversation. I'm always appreciate when really smart people can explain things without getting too technical.
I feel this conversation is one of those things that you didn't really know you needed until you got it.
wow! this is really awesome! keep up with the good work. you are really inpiring us who have limited resources to get such information in time
Thank you for adding value to my life. Your podcasts are really helpful.
I love that there's a dedicated section to Jeremy Howard. Absolute legend that guy!
I have a feeling this guy also knows Ada. First language I learned and it seems like he's taking a lesson from both parties and making the best of it.
enjoyed the whole thing so much. I'm so with Chris on complexity being THE enemy ... took me a long time to learn. Python was my first love and I still love it. However since I found Erlang/OTP and the BEAM runtime I've come to believe this is the strongest programming env and runtime in a surprisingly large number of domains. especially when combined with a language like Elixir which has been designed for similar goals as what Guido had in mind for Python. The fundamental message-passing concurrency, the error handling philosophy, and now even the ML capabilities with NX and higher-level libraries. Joe Armstrong's thesis made me realize how much of a secret sauce we've got in our hands.
As Chris days "when everyone goes left, you sometimes have to go right". Python might be the clear winner in the AI/ML space for now .... but I'm betting it might not be forever 😉
Thanks Alex, it is intriguing,educating and very instructive! 75 K views in 12 hours ! Congrats
You should have monthly Chris episode (and couple others :) )
One of the few podcasts where you can have very different "auto tune" discussions with Ye and Chris Lattner
CL is just wow. Thanks for Part 3. When Mojo was announced I knew it was just a matter of time before Lex had him on again.
Lex, you gotta have this guy back now, a year later, to get an update. I've been coding in Mojo and I friggin love it.
Chris is a legend, such a humble person.
Great talk, unfortunately he misinterprets the "zero cost" meaning in "zero cost exceptions" (and it's *not* zero cost exception *handling*), the same way many people misinterpret the term zero cost abstractions. The zero cost refers to the fact that you don't have any additional runtime cost, specifically in the happy-path, meaning that if a function doesn't throw, there's no downside of having the `throw` in the function itself from a performance perspective. But ofc depending on your definition of "cost" you'll still get larger binaries, require RTTI and so on, but that's usually not what people find relevant in 99.9% of cases. It's the exact same thing regarding zero cost abstractions, you have tons of abstractions that don't affect the runtime, but ofc it'll affect lesser things such as compile times and what not.
And talking about returning values that represent errors like you do in Rust, or with `std::expected` since C++23, that's not relatable to exceptions. Exceptions and result types cover different areas of handling errors that aren't necessarily interchangeable, hence having the support for both is optimal.
Heard about mojo from Fireship. Great start to the weekend!
he make a good point about the file extension and looking at them, took me a week or two to remember the order of .ipynb properly. And I've very stoked to see more from mojo.
Your discussions with Chris are always such a delight ❤
I loved any minute of it. If someone can solve this complex AI problem, then it is most probable Chris. Playing around with Mojo from its infant stages feels like being part of the history. Thank you, Lex! Thank you, Chris!
Молодец Лёша! Благодаря тебе у нас есть возможность увидеть и услышать «особенных» людей. Chris seems to be such a humble guy, great interview!
I love how excited Lex is for the conversation. He's a programming nerd like me. This stuff is fascinating even though I don't fully understand everything...yet.
Three hours later, still no idea how Mojo unifies things.
It’s a scaling approach- factor, factor, factor!! Lol
The main selling point for mojo is that you can leverage MLIR to write custom operators instead of relying on a runtime written in C++. Like for tensor operations pytorch uses Aten as the runtime which is written in C++. Mojo is like pytorch GLOW or tensroflows JAX. I don't understand the need for mojo when there are better solutions.
@@solitary_crow I think they are trying to be TypeScript for AI. I have the impression that Chris saw a bussiness opportunity for LLVM like stacks in proliferation of hardware in AI and decided to create such a stack with modular. As Python is the language of AI, they chose Python as an interface to their stack in order to attract users.
Having chris lattner on your podcast for the second time make me smile happy. James Gosling also genius person have so many wise and knowledgable experiences in the IT field. could you invite him for the second round lex?
Ok
packaging in python with things like poetry is pretty straight forward and clean, rarely problems with c or c++ dependencies (this can happen on certain platforms like windows where things might not be well tested, or new hardware like apple silicone where the compiling process might not be optimal or correct)
The Dream Team of programming. The greatest programmers of all time and Chris Lattner.
This talk was fantastic. I’m not a programmer, just someone who has been using and navigating Linux. Would this language be a good language to start?
No, start with python.
You are awesome lex and also your guests. It's entertainment mixed with education.
Would love it if Apple embraced Mojo.
What a great interview! I love the chemistry, clearly they enjoyed this discussion:)
Phenomenal video, fell in love with hardware as a child and became a software guy as an adult, seeing it all morph into these heterogeneous systems is magical
Can someone provide any resources to become hardware literate? As a data scientist, I'm intrigued by the intersection of software/AI and hardware (which is mainly what Chris Lattner talks about), but I find it quite challenging to find sources that can help me become sufficiently literate without becoming an expert (I haven't found any course from Chris Lattner or similar, it would be nice if there is one haha)
Lex you know we gonna be hanging on right here even when we get lost.
mojo gang where u at 👀
Very interesting video.
Since Chris mentioned Zig, I see a lot of parallels in the way they are operating. Zig seems to be following that Swift model to bring the C community along, while Mojo looks to be using the release early and get feedback open model Zig is using.
I wonder if they have considered using Zig to solve the C packaging and compiling portion for python to help facilitate the transition? After realizing this problem, Zig seems like a natural fit for Python and Mojo for this.
Wish Lex would have more devs on for us ❤️
So informative, thanks Lex!
This is amazing! Chris a legend.
Amazing. Ive been comissioning computers on site for years now and i have been watching operating systems, software, everything become more bloated, obnoxious and convoluted to use. Faster computers do indeed make for lazy programmers... Chris is the first programmer ive seen who 'gets' the horrendous inefficiencies that we have learned to put up with with our hyper fast computers - to hear he has made a language thats 35,000* faster than everyone else... It makes me think about american cars in the 60's - fuel was so cheap it was easier to just pout a bigger engine in if you wanted to go fast. Nowadays a 1L motorbike engine can make the same power as the 7.2L v8's of old because of the constraints, fuel pricing etc. We would have come up with this language ages ago if we didnt have moores law. Keep it up Chris i was just about to jump into learning python, looks like mojo now!
Chris: package distribution, compiler interface design, let/var....
Lex: I hear you... what is the meaning of life?
Watching this episode with ORTUS is something else
Ty for applying it Lex!
Please ask every guest you have for advice for young people. It is the most important segment that I always look for in a video. Please ask for advice for young people from every guest.
Oh man, you are my hero. I got a lot of knowledge from this interesting podcast.👍
That intro gave me chills 🔥
22:45 I worked in an R&D lab and we described our work as mainly S&R, Search and Redevelopment.
Hey Keith it’s me Ryan from the office!
@@boi__7898 HI Ryan how's it going
I would like to hear Chris' thoughts on where Modular stands relative to ONNX/ONNX Runtime.
Please round 2 with Jeremy Howard!!
finally been waiting on this for a while lex😃
Love how you mention list comprehension with the things you love about python. Ever since I learned how to write them I never populated lists in the same way. I use them almost to a flaw lol
The more I listen to that podcast the more I realise how much I don't know and how clever some people are
Fascinating from begin() to end()! I can see that Chris would be a great person to work with. If I were looking, I would drop him a line. I hope Modular does well.
___ Don''t do drugs
Everyone, even non-programmers, should listen to the last 30-40 minutes.
Always fascinated as to the progress of the Human condition. You inspire me more with every new podcast. If we are to advance, it will be with the optimization of our marriage with AI, and even like a marriage things will be tense but the outcome will be forever positive.
Great episode, just signed up for Mojo early access
Just finished listening to this episode over multiple sittings. Amazing content packed with fascinating insights, 120% must-see for anyone who is interested in the evolution of programming languages. Thanks so much for doing this!
I haven't programmed anything since Basic64 and I did not understand 5 minutes of this podcast. It was still pretty good tho 😂
What a great conversation! Thank you Lex!
Thumbs up before listening to the Podcast. I love Chris Lattner and Lex
Truly a brilliant mind, and an awesome interview!