I think abstraction and improperly preplanning your program's structure are two completely different things.

@@DavidsKanal I was of course talking about unnecessary abstraction and it then imposing a bad structure on the code. And so far I've yet to see a "proper" way to preplan software before some form of prototype and understanding of the problem exists.

Yes, I agree. It's doing speculative programming, way ahead of any concrete solution exploration of the issue. Waste of time, and likely wrong. People get sucked into systems and design patterns, rather than focusing on the actual solvable problem, and that cost is ridiculous. People used to write complex software is shorter time periods than today, with more reliable results. Jon mentions doing the simplest and most direct thing, and then refactoring later if more is needed. He's absolutely correct. The other method that most seem to do these days is backwards, and ends up in a bloated mess that is difficult to refactor, because the "solution" was more about patterns and abstraction, rather than the fundamental problem that is being ultimately tackled.

@@RockTo11 Well, I agree with you. Personally, I never even fell into this pattern of speculative programming - it makes no sense to me. If having finished multiple projects taught me anything, is that you'll end up refactoring lots of stuff 5+ times anyway - it's a waste of time and energy to preplan stuff you can't predict. I don't get how more people see this. It's painfully obvious if you've ever programmed anything non-trivial; at least to me.

Mike Tyson said it best: Everyone has a plan until you get punched in the face. I remember writing code for a disk controller and got my ass whipped. It was well structured and abstracted to the point of supporting other disk mechanisms too. With a simple script and a configuration database, I could generate and compile code for almost any drive made. It worked, but the performance was dismal. It just banged out bits and polled for results. To make it faster, I had to use lots interrupts, which turned my masterpiece into Spaghetti code and universal support impractical.

I saw your comment on John carmack interview and you mentioned John blow and then I started watching him, and now found you here lmao

@@jazzgirl2358 Haha Yeah listening to the Carmack interview made me want to go back to these old Jon clips and see if my position on them changed after a few years. Looking back at it, it is kinda cool how similar John and Jon's view seems to be on the matter.

Which John Carmack interview? Just making sure I didn’t miss one.

@@FastFSharp The Lex Fridman one a month or so ago.

C, at least for me, was easy until I got into multithreading, but even excluding that, it was a bit of a headache sometimes

Yeah, I started programming Java in the late 90's and fell hard for all the OOP stuff. It took years to unlearn all that dogma.

Why should we expect to ever get there? All attempts to simplify programming so far have ended in complexity disasters.

I think the problem is actually deeper, in that our society has an education and intelligence problem. Lower standards in schools, degenerating culture, etc. Knowledge and the capacity for knowledge is being lost in many fields.

@@___Hermitage I agree. All of this assumes high quality human capital, but does that actually exist in the amounts needed?

@@HairyPixels I don't think it ever existed. Most people are and always were just OK at their jobs. This is how statistics work. Most construction companies can build a detached home. Vey few can build a skyscraper. Fewer can build either well. We've been building houses for thousands of years.

@@FicoosBangaly The quality of the NEW programmers has been going down due to the lowering of entry requirements. Back when I started to program in 1997 you were forced to learn yourself from first principles but nowadays you can just watch tutorials on KZfaq.

@@DemonixTB chill, I’m not telling anyone what to do… just pointing out

@@DemonixTB but that is a good point, creating a language is not solving a problem, the same way creating an abstraction is not solving a problem, they are both tools that help you solve problems later. and trust me, a bad language is WAY worse than a bad abstraction and it takes WAY more time to make.

@@MsBukke Yeah, I think that's why he didn't just release it publicly early. The future will show whether Jai is worth it, but I definitely think we are past due for a better C-like language. And Rust isn't necessarily the right solution, IMO, at least not for everyone. Zig, Odin, Nim, Jai, seem to be poking at this from slightly different directions, and it's good that people are trying. I hope in the future I can reasonably write code for projects where currently I need to use C++, but in a better language.

@@davidandrewthomas It's pretty amazing how Jai sort of kickstarted people to create languages for solving these kinds of problems, can't wait for it to get released :)

@@DemonixTB pour out the koolaid. There are low level languages that you can do that perfectly fine in. C++, Rust both are great examples. C is alright, but writing in it is unergonomic. 75% of what we write is trivial stuff and C makes writing that verbose thus wastes our time. There are so many things that already waste our time, bugs, logic errors, etc, that we cause ourselves. You can write C style code in Rust or C++ or any of the other systems programming languages. Jai is going to bomb, I am pretty certain of that. But if it doesn't, cool, we got another great language if I am wrong.

it needs to be so to make up for the jobs stolen by AI and other automation

​@@mitnick212 AI really hasn't stolen any jobs. The modern AI application to problems are typically horrendous enough that most companies have thrown away billions of dollars into data research that do not even yield significant real world solutions since 2011. AI is a VC term utilized for other companies to signal to investors to throw capital at that company until they burn out and eventually sell themselves to Google, Facebook or Amazon

if I understand correctly, WebAssembly enable web apps run near OS level, right ?

@@mitnick212 There's significant overhead and delay I believe, when communicating the results with the website/webapp itself

Something better than the web browser to host apps? Are you suggesting we should write code that runs on the user's computer?! Come on, that's such a 201x thing to suggest 😏

@@ilonachan Only if writing such code wasn't such a shit experience..

It's unrthodox perhaps but I like to view Objects not as representatives of something to be modeled as they teach you in OOP 101, but rather try to design them as essentially programs or subsystems in itself that combine with different handles to make up the complete software. Think of it as an engine you drop in your car or a shell tool you install into a Unix system. It avoids the pitfall of stack trace spaghetti while still providing some level of modularity to the design.

@@PixLgams if it works for you and your users, great. the only thing I'll mention is that what you're talking about sort of sounds like a module system. i think just having explicit support for modules is a better idea. but, at the end of the day it's all about the end product.

@@PixLgams Yes, like Warren said, what you're describing is 'modular programming'. This is almost always the mistake that OOP coders make -- that is to say, to use crude objects as modules. What you ideally want is a program made up of modules that are only connected to the rest of the system via one function (or as few as possible, usually init() and uninit()). That one init() function could take 2-50 parameters, including function pointers to things that it needs, such as a malloc() function. This way, when you want to replace the module, or when you've replaced the allocation module, you don't have to change the 2000 lines of code that use the allocation function that you reference. In C, you would use static on all your functions and 'global' variables (This way they're only 'global' to this file), except the init() and deinit(), and as few other funcs as possible. This is proper encapsulation. Look at the Quake 2 source code for examples of minimalist interfaces (esp the gfx interface) in pure C, in a data-driven paradigm.

OOP has nothing to do with modules, objects it's about pets - dogs, cats and even chickens. I know I read 20 books about it.

@@PixLgams That's fine and all, but this already existed decades before OOP in the form of modules,. APIs, library interfaces, etc.

CSS is fine-ish for the domain it is designed for. Also a graphic interface is THE ONE example in which objects make 100% sense, since the UI abstraction is literally made of objects.

Procedural style is my favorite, sometimes though you can't get away without abstractions. But they should come afterwards, not on the initial stage.

Yes but quality considerations only think about passes but never speed. Is it really quality for something to take 100s when it could take 0.1s?

@@oblivion_2852 thats not correct. Tests can measure performance aspects as well. And yes, Some check correctness, some memory, some check errors and so on dependong on the product. Tests are 10% for verifying content and 90% making sure they dont break in the future.

I don't think anyone is against abstraction per se. What Jon is pretty much saying is that if you do a cost-benefit analysis of your abstractions then you will find out that a large number of them are not worth it.

This post is the dumbest thing I've read all day

@@Cons-Cat no, he is a 100% correct. Without it, there would be no malloc, no alloca, no C standard library. All abstractions. But I suppose you've written your own std lib? And if you have; congratulations, you've written abstractions.

​@@simonfarre4907 he's a low level system programmer, he obviously knows those are abstractions. He's not telling anyone to write in assembly what he's saying is that OOP abstraction are mostly bad/widely misused. He's arguing against the implementation not the concept.

Same can be said about any modern industry, not just software engineering.

Issue with object-oriented as an idea is that it's terrible for modern processors. What you need for code to run fast is for most things below a certain level to run tight loops on packed data. Object-oriented ideas really fight that, unless it's used as a module system basically at the periphery of an application to wire components together.

@@davidandrewthomas well yeah, I wasnt saying that true OO is better than the current definition or anything I was just sayibg that the definition changed.

@@jac1011 yeah, that's definitely true.

it still sucks tho... and is useless and bad

@@KookoCraft Nah it's fine. Just another tool. Blow's a dummy.

OOP is a monumental fucking disaster for large scale projects. The only successes I've seen basically had to rewrite large portions of the code base in a style that mostly bypassed all of OOP's supposed strengths. That ends up being the only way to be able to make changes that don't feel like trying to swim in concrete. Funniest thing I've ever seen was in a very large web application with a really impressive 500,000 lines of code class hierarchy that developers got so frustrated with that they replaced it with a single class that ONLY funneled name/value strings from the database to the web page, and a couple of other classes handling security checks (SQL injection, etc). That actually saved the company from going under, as they could now give customers what they wanted within a two-week development cycle instead of 6 months.

The issue is that nobody can agree what OOP's "best practices" are suppose to be and what is misused or overuse, because its solutions are proposed not on measurable results but subjective philosophical concepts.

The irony is that the "bad" OOP code is created by the people trying to "do it right" and follow everything OOP gurus say by the letter. The "good" (acceptable) OOP code is create by the people who don't give a damn and just do the minimum amount of OOP to pass the code review.

With other highly specific criteria.

@championchap I can assure you that when my 15 year old says "meme", he's not referencing The Selfish Gene. I did explain the concept to him once, he's probably forgotten.

No, that is how entropy works

@FichDich InDemArsch Why the 🤦‍♀️??? This is a sophomoric, but valid point everyone here brings up. From my experience, it's the problem solving framework (rather, a lack of a good framework) that's the biggest problem across several industry, not exactly the languages that are used. This would be why corpos value problem sustaining practices and capital over problem solving.

@FichDich InDemArsch I don't think the op was saying that there's nefarious intent at play. Moreso that there are selection pressures in the industry that reward being first to market/shipping things quickly. This results in people using tools and processes that make development time fast, possibly at the expense of the end product. This obviously isn't the only reason for the problems in SE, but it is definitely a cause of some of the problems.

@FichDich InDemArsch Measure/profile all the things: from development/debug time to execution/dev comprehension time with control variables just like you would with any scientific experiment for an empirical study. Inlining code before abstracting the commonalities into pure functions is one of the many techniques that helps you converge towards the "nirvana" between genericity, expressivity, and performance, which is what all code should strive for. Knowing what you know, knowing what you _might_ know, knowing what you actually don't know is essential for the design of a solution (See that Symmetry and Orthogonality talk by Charley Bay for other gems related to this); though it might take some experience to create mental shortcuts made after seeing certain problem shapes. Thankfully there are plenty of PKMs like Obsidian and Roam to help out with storing mental shortcuts and most common searches.

@FichDich InDemArsch Emotional empathy to find use cases and counter-examples for the knowledge banks like I mentioned before helps with the bottom-up software architecture. Finding "general-enough" patterns/concepts that help with the top-down software implementation. There's also this idea of using incremental development to improve development processes/feature sets and iterative development to improve product quality simultaneously (with reflection) as noted by Alistair C0ckburn. Also have a paper indexable commonplace book with consolidated fundamental truths that can be applied to a number of sources.

@FichDich InDemArsch Appreciate your latest response. I agree that there's a disjoint as far as what "good code" is. Having such a meta industry like "software engineering" that is considered an "art" first which encompasses many different topics explains why there's little overlap across all industries as far as what best practices apply universally. Doesn't mean they can't be written down in some central authority looking over all of software engineering. With the current science community in govt that focuses more on narratives over "scientific integrity", looks like we all have to establish those rules/guidelines together from the bottom up before they do. I hope that empirical studies are conducted in the near future to establish those "good practices" in industry and education.

A language, then a couple of games before (maybe) getting to an editor, OS, and terminal emulator?

That would be the most cynical possible interpretation, yeah.

If there was a better language to write his own language then he wouldn't write his own language. He would just use that better language. C++ being the best we got is the tragedy. Thats my interpretation.

Bjarne did say he wanted his language to be general and designed to evolve. Clearly there was a disconnect between his ideas and those of "the committee". If only they took his ideas of "concepts" earlier on, then generic meta-programming would be so much easier than it is now.

@@krinjon193 Why doesn't he use C

@@krinjon193 the better language is C

@WS WS If it has autocomplete, I consider it IDE. Therefore the terminology may be inconsistent between me and others.

@@AN-ke2uz Well I've been working as a software engineer for nine years and it never caused me any trouble. Just a tip: Be careful when using the word "stupid". People may get offended because they would think that you are calling them dumb.

I've been using VSCode since it pretty much came out. At first the autocomplete worked fairly well. But over time it has started failing more and more (not loading the suggestions in time, not finding the suggestions etc.). Now it fails many times a day for more, causing a lot of annoyance and swearing. If I were a bigger man I would work to undo my autocomplete habits. I can see how Jon with his 0 tolerance for bullshit would not accept something like this.

"If it has autocomplete it is an IDE" Gmail has autocomplete Ergo Gmail is an IDE

@@TypingHazard A guy killing cockroaches is a debugger. Therefore he is also an IDE. Yes, I still defend my position, thank you for asking.

you're one of those guys that throws accusations of elitism and presumption every time someone talks about anything deep and provoking. He's attacking the methodology and philosophy used by the industry but you just gotta to make it personal and feel like it's some attack towards other people's intelligence/professionalism.

law kills everything

go back to basic(s) pun intended

when the language became the problem, it was lost

when pride of the methods became the problem

someone needed to be paid

How can you really say that? Don't you constantly criticize game software for being an unoptimized mess?

You're in a dream land. I am a civil engineer in bridge design. Aside from all the other information of 44 years of living, I watch a lot of videos on line on different things, and I can tell you there is a consistent pattern running though everything. The core problem is not in programming it is in our society. Things are not continuously improving, they are running downhill fast. There may this and that that is improving, but overall things are not. There is a an almost irresistible tendency toward making things overly complex in every direction. At the same time it takes more and more people to accomplish even less. There is too much abstraction. Too much high mindedness. Too many people trying to prove how smart they are. Too many people who do not want to take the time to understand things in a deep way. To think is not the same thing as to do. Too many thinkers are trying to tell the doers how to do.

Lol

@@adrianalanbennett we all know how it feels like to read code made by a guy trying to show how smart he is..... I feel your pain.

Facts !!!