We got to build mini versions of the first 4 languages (imperative, lisp, ML, Smalltalk) in the PL course at tufts which is now published as a textbook [1]. There used to be a prolog part that sadly got cut.
One correction I'd make to the article's taxonomy: Ruby is an object oriented language not an Algol. Its inspiration is Smalltalk, and much of the standard library naming comes from that route (eg collect rather than map).
Ruby is object oriented from the ground up. Everything (and I do mean everything) is an object, and method call is conceived as passing messages to objects.
While Ruby is most often compared to Python (an Algol), they come from very different evolutionary routes, and have converged towards the same point in the ecosystem. I think of Ruby as a cuddly Alpaca compared to Python's spitting camel.
Since Python introduced new style classes, it also became a pure OOP language, even though it might not look like it at "Hello World" level, all primitive types have become objects as well.
I have found the definition of OOP to be fuzzy. For example, I don't see why having methods would make a data type object oriented. I associate OOP with factories, inheritance, using classes in places that might be functions otherwise, and similar abstractions.
Perhaps this is the counterfactual: I program in Python regularly, but don't program in an OOP style; I use dataclasses and enums as the basis, in a way similar to Rust, which by some definitions can't do OOP. So, if Rust can't do OOP (assumption) and I can write Python and Rust with equivalent structure (Assumption), does that mean Python isn't strictly OOP?
Rust can definitely do OOP, not only I could easily convert "Raytracing in one Weekend" from C++ into Rust, while keeping the same design, Microsoft has no issues adding a Rust projection for COM/WinRT, which are OOP ABIs.
However that is not the same as Python, which is 100% OOP, when using basic stuff like numbers, from Python language semantics those things are object, and there is the whole machinery in place even for basic stuff like addition.
And yes OOP is fuzzy from CS point of view there are multiple approaches and it doesn't get reduced to the way C++ and Java do it, just like FP and LP are fuzzy as well.
Some folks would swear if it isn't Haskell or Prolog, than it isn't FP or LP, when a CS book and programming language evolution will be more fuzzy.
That seems like a pretty lame gotcha--saying "Aha! The language you write in uses your hated paradigm under the hood" seems to invite the immediate response of "So? I don't use it."
I think the choice to identify a specific ur-language as "Object oriented" throws people off since OO is just a style of programming in the same way that procedural is. I don't think it's useful to say that Python and C++ are both the same kind of language because they both have multiple inheritance, rather that's just an observable commonality, like noticing that both Delhi and Vegas are too hot. Yeah, but I don't think that's because they're the same kind of place...
Object-oriented programming is less of a syntax and more of a philosophy. While Erlang’s syntax belongs to the ML family, its creator, Joe Armstrong, argued that Erlang aligns more closely with the original OOP philosophy defined by the inventor of Smalltalk than Java does: 'The essence of object-oriented programming is messaging, not classes and inheritance.'
I might add another class of languages: those intended to express proofs, via the Curry-Howard correspondence. Lean is a primary example here. This could be considered a subclass of functional languages but it might be different enough to warrant a separate class. In particular, the purpose of these programs is to be checked; execution is only secondary.
> Agda, Idris, etc. are functional languages extended with complex types
I think they are not. No amount of type level extensions can turn a regular functional language like Haskell into something suitable for theorem proving. Adding dependent types to Haskell, for example, doesn't suffice. To build a theorem prover you need to take away some capability (namely, the ability to do general recursion - the base language must be total and can't be Turing complete), not add new capabilities. In Haskell everything can be "undefined" which means that you can prove everything (even things that are supposed to be false).
There's some ways you can recover Turing completeness in theorem provers. You can use effects like in F* (non-termination can be an effect). You can separate terms that can be used in proofs (those must be total) from terms that can only be used in computations (those can be Turing complete), like in Lean. But still, you need the base terms to be total, your logic is done in the fragment that isn't Turing complete, everything else depends on it.
This is just wrong, you're being too didactic. Idris specifically lets you implement nontotal functions in the same way that Rust lets you write memory-unsafe code. The idea is you isolate it to the part of the program with effects (including the main loop), which the compiler can't verify anyway, and leave the total formally verified stuff to the business logic. Anything that's marked as total is proven safe, so you only need to worry about a few ugly bits; just like unsafe Rust.
Idris absolutely is a general-purpose functional language in the ML family. It is Haskell, but boosted with dependent types.
Random passerby chiming in: so this means you can write "regular" software with this stuff?
While reading TFA I thought the theorem stuff deserved its own category, but I guess it's a specialization within an ur-family (several), rather than its own family?
It definitely sounds like it deserves its own category of programming language, though. The same way Lojban has ancestry in many natural languages but is very much doing its own thing.
Yes Idris was meant to write regular code. F* is also meant to write regular code
But I think that the theorem prover that excels most at regular code is actually Lean. The reason I think that is because Lean has a growing community, or at least is growing much faster than other similar languages, and for regular code you really need a healthy ecosystem of libraries and stuff.
> if you're a developer who wants to exploit the multiplicative factor of a truly flexible and extensible programming language with state of the art features from the cutting-edge of PL research, then maybe give Lean a whirl!
Does not sound that appealing to me. Sounds like little consistency and having to learn a new language for every project.
> You can separate terms that can be used in proofs (those must be total) from terms that can only be used in computations (those can be Turing complete), like in Lean
What I meant is that the part of Idris that lets people prove theorems is the non-total part
But, I think you are right. Haskell could go there by adding a keyword to mark total functions, rather than marking nontotal functions like Idris does. It's otherwise very similar languages
> To build a theorem prover you need to take away some capability (namely, the ability to do general recursion - the base language must be total and can't be Turing complete), not add new capabilities. In Haskell everything can be "undefined" which means that you can prove everything (even things that are supposed to be false).
Despite what the fanatical constructivists (as opposed to the ones who simply think it's pragmatically nice) seem to want us to think, it turns out that you can prove interesting things with LEM (AKA call/cc) and classical logic.
I mean you are kinda right but kinda wrong. To get a proof checker you take a typed lambda calculus and extend it with Pi, Sigma, and Indexed Inductive types while maintaining soundness.
Yes haskell's `bottom` breaks soundness, but that doesn't mean you need to take away some capability from the language. You just need to extend the language with a totality checker for the new proof fragment.
Lean is definitely a dependently typed ML-family language like Agda and Idris, so "ML" has it covered. And the long-term goal of Lean certainly is not "execution is only secondary"; Microsoft is clearly interested in writing real software with it: https://lean-lang.org/functional_programming_in_lean/Program...
OTOH if you really want to emphasize "intended to express proofs" then surely Prolog has that covered, so Lean can be seen as half ML, half Prolog. From this view, the Curry-Howard correspondence is just an implementation detail about choosing a particular computational approach to logic.
These are not true programming languages because by definition they are not Turing complete. If they were Turing complete it would be possible to write a false proof that just compiled down to a non-terminating program.
I feel like Turing completeness has always been set as the boundary of programming language if there's any boundary at all. That's what people has been using to not include HTML as programming language for example. Or to include MTG as one.
I think it's a pretty recent thing. Turing completeness is neither a sufficient nor a required property.
HTML is not a programming language, it's a markup language - it's in the name, and it's the way it is used (it's not used to describe any kind of computation, it's straight up data that is parsed by algorithms).
Neither is PowerPoint, or game of life a programming language even though both are Turing complete.
Turing completeness is the upper bound of computability, not the lower bound. It's useful mostly for showing that some thing can express the full range of computable problems, or for snarking that some thing is far more complex than it has any right to be.
Total languages omit partiality and non-termination from Turing completeness.
Partiality is IMO irrelevant when it comes to computability. Any partial function (that is, one whose range is not defined over its whole domain) can be expressed as a total function by either constricting the domain or expanding the range. For example, a "pop" operation on a stack is not defined for an empty stack. You can just loop forever if pop() is called on an empty stack. Alternatively, you can require that pop() is given a witness that the stack is non-empty, or you can require that pop() returns either the top-most element of the stack or a value that indicates the stack was empty. Both let you compute the same set of things as the former.
Non-termination is required to be Turing complete, because being Turing complete means being able to compute functions that one cannot reasonably expect to complete before the heat death of the universe. In _practice_ every function terminates when the computing process dies due to some external factor: process runs out of memory ("real" Turing machines have infinite memory!), user runs out of patience, machine runs out of power, universe runs out of stars, that sort of thing, so _in practice_ doing 2^64 iterations before giving up will generally* give you the same outcome as doing an unbounded number of iterations: it'll either terminate, or the process will be killed (here, due to reaching its iteration limit).
On the flip side, giving up non-termination and partiality only gives you increased correctness. If there's one thing we've definitely established in computing, it's that we will readily discard correctness to gain a little extra productivity. Why make a developer implement code to handle reaching an iteration limit when you can just make the user get sick of waiting and kill your app?
* 18 quintillion is a very large number. Have a try. The most trivial recursive function, on my M4 Mac, when convincing clang to be smart enough to turn it into a loop but dumb enough not to elide it altogether, would take a bit shy of 600 years to complete if iterating ULONG_MAX times; I didn't wait for that, if I'm honest with you, I ran it with a much smaller iteration count and multiplied it out.
I recently revisited a language comparison project, a specific benchmark tallying the cycle decompositions in parallel of the 3,715,891,200 signed permutations on 10 letters. I kept a dozen languages as finalists, different philosophies but all choices I could imagine making for my research programming. Rather than "ur" I was looking for best modern realizations of various paradigms. And while I measured performance I also considered ease of AI help, and my willingness to review and think in the code. I worked hard to optimize each language, a form of tourism made possible by AI.
Naively this is quite surprising, but the devil is in the details. With the exception of Lean I'd point out they're all fairly close: Chez being 2.5x slower than C++ is not ignorable but it's also quite good for a dynamically-typed JITted language[1]. And I'm not surprised that F# does so well at this particular task. Without looking into it more closely, this seems to be a story about F# on .NET Core having the most mature and painless out-of-the-box parallelism of these languages. I assume this is elapsed time, it would be interesting to see a breakdown of CPU time.
I don't think these results are quite comparable because of slightly differing parallelism strategies; I'd expect the F# implementation of just spinning off threads to be more a little more performant than a Rayon parallel iterator, which presumably has some overhead. But that really just shows how hard it is to do a cross-language comparison; Rust and C++ can certainly be made faster than the F# code by carefully manipulating a ton of low-level OS concurrency primitives. This would arguably also be little misleading. Likewise Chez and Haskell have good C FFI; does that count? It's a tricky and highly qualitative analysis.
[1] FYI, one possible performance improvement with the Chez code is keeping the permutations in fxvectors and replace math operations with the fixnum-specific equivalent - this tells the compiler/interpreter that the data are guaranteed to be machine integers rather than bigints, so they aren't boxed/unboxed. I am not sure without running it myself, but there seems to be avoidable allocations in the Chez implementation. https://cisco.github.io/ChezScheme/csug/objects.html#./objec...
Thank you. I will try your Chez idea. I love Chez, even if coding in Scheme can feel like rubbing sticks together to start a fire on an island, when e.g. Scala has induction ranges. And I didn't try Idris or Racket as they compile to Chez, but perhaps they do so better than I did.
As for parallelism this is a primary concern of mine, and I tried multiple approaches for every language where there was a choice. I used my own work-stealing code only when it beat standard libraries. AI warned me I was in over my head, that writing such a library takes years of experience, but my use case (and my expected use cases in my research) is so uniform that simple can win, minimally touching the required bases such as permuting tasks to avoid false sharing.
I don't believe that the JIT languages (F# on top) do so well because of better parallelism. This is branch optimization. For this use case an AOT compiler with ample benchmark data to influence output should do better. That isn't a thing, and the argument seems to be that few use cases stay consistent. A JIT can adapt.
Yeah :/ For a larger program you can pay the readability toll once, via a syntactic form that expands the general vector/arithmetic operations to the fixnum versions, e.g. used something like
(define (heap-permute! perm j callback)
(with-context 'fixnum ;; same trick works with 'flonum for 64-bit floats
(let ([n (length perm)]) ;; actually fxvector-length
(let generate ([k (- n 1)]) ;; actually fx-
(if (< k j) ;; fx<
(callback perm)
(begin
(generate (- k 1)) ;; fx-
(do ([i j (+ i 1)]) ;; fx+
((>= i k)) ;; fx>=
(if (even? (- j k)) ;; fxeven?, fx-
(swap perm j k)
(swap perm i k))
(generate (- k 1)))))))) ;; fx-
Sorry if I borked the indentation. I have been working on stuff like this, and more general macros around dependency injection and inversion of control (e.g. you could write this macro to take the type as a parameter and generate code optimized for 'bigint or 'rational). Maybe check back after the summer :)
And BTW I misspoke earlier, of course Chez is AOT rather than JIT. From one approach it's sort of a hybrid: really fast on-the-fly AOT kinda looks like JIT, tongue-in-cheek you could say "NoT compilation" (nick-of-time). But proper JIT of course has huge advantages. If you reeaaaallly wanted to sabotage readability, Chez makes it easy to invoke the compiler at runtime, so along with the C FFI I think you could hack together some sort of JIT. But wow, what a mess that would be! You'd better be getting a PhD thesis out of it :) And if the performance is that critical you'd be much better off with F#.
I haven't looked into the code, but Lean being so slow may be misleading depending on how you benchmarked it. IMO the fairest test is how "Lean code" (or Rocq code, etc.) is actually run, which is as native C code following extraction.
Given the sane C defaults that are applied by code extraction techniques, the delta really shouldn't be so great. But it's a common pitfall to torture one's own verified code in order to get it proven, and I'm also not sure how good of support there is for parallelism.
there's a few more semantic families: verilog, petri nets and variants, Kahn process networks and dataflow machines, process calculi, reactive, term rewriting, constraint solvers/theorem provers (not the same with Prolog), probabilistic programming,
plus up and coming (actual production-ready) languages that don't fit perfectly in the 7 categories: unison, darklang, temporal dataflow, DBSP
It may feel like a little bit of cheating mentioning the above ones, as most are parallel to the regular von Neumann machine setup, but was meaning for a while to do an article with 'all ways we know how to compute (beyond von Neumann)'.
> was meaning for a while to do an article with 'all ways we know how to compute (beyond von Neumann)'.
Would be very glad to read this.
In the meantime, I reproduce a part of an article by Steve Yegge:
---
What Computers Really Are
Another realization I had while reading the book is that just about every course I took in my CS degree was either invented by Johnny von Neumann, or it's building on his work in mostly unintelligent ways.
Where to start? Before von Neumann, the only electronic computing devices were calculators. He invented the modern computer, effectively simulating a Universal Turing Machine because he felt a sequential device would be cheaper and faster to manufacture than a parallel one. I'd say at least 80% of what we learned in our undergrad machine-architecture course was straight out of his first report on designing a programmable computer. It really hasn't changed much.
He created a sequential-instruction device with a fast calculation unit but limited memory and slow data transfer (known as the infamous "von Neumann bottleneck", as if he's somehow responsible for everyone else being too stupid in the past 60 years to come up with something better. In fact, Johnny was well on his way to coming up with a working parallel computer based on neuron-like cellular automata; he probably would have had one in production by 1965 if he hadn't tragically died of cancer in 1957, at age 54.)
Von Neumann knew well the limitations of his sequential computer, but needed to solve real problems with it, so he invented everything you'd need to do so: encoding machine instructions as numbers, fixed-point arithmetic, conditional branching, iteration and program flow control, subroutines, debugging and error checking (both hardware and software), algorithms for converting binary to decimal and back, and mathematical and logical systems for modelling problems so they could be solved (or approximated) on his computing machine.
Von Neumann may possibly have been the smartest man to ever live, but giving him credit for all of this is too much, brushing aside many other inventors (oft independent, to his credit).
I volunteered to do the formula parser, thinking it sounded like a fun challenge.
I was stumped for a week, until I realized I could rewrite the formulas into a form I knew how to parse. So it would rewrite 1+1 into ADD(1,1) and so on.
I also refused to learn regex, so the parsing code was "interesting" ;)
I recall a comment from a colleague. "Okay, Andy says it works. Don't touch it." XD
Guy from another group used regex and his solution was 20x shorter than mine.
Regular expressions are probably not enough for parsing formulas (depending of course on the exact task given), they usually are at least a context free language.
Regular expressions are definitely enough for turning characters into tokens, after which a simple recursive descent parser is vastly more straightforward to write. Lexing is optional, but generally advised.
Plus up and coming (not quite production-ready IMO, but used in production anyways): ChatGPT and the like.
Of course, it’s debatable whether they are programming languages, but why wouldn’t they be. They aren’t deterministic, but I don’t think that is a must for a programming language, and they are used to let humans tell computers what to do.
You can just get the code without buying the book, learn with Simply Scheme or any other
book and apply the functions from the code, the solvers are really easy to understand.
Great list of languages that don't fit the conventional families. I've been curious about some of them, like Petri nets and term rewriting, and will enjoy exploring the others.
Found a working link to the paper about propagators.
We agree on Algol, Lisp, Forth, APL, and Prolog. For ground-breaking functional language, I have SASL (St Andrews Static Language), which (just) predates ML, and for object oriented language, I have Smalltalk (which predates Self).
I also include Fortran, COBOL, SNOBOL (string processing), and Prograph (visual dataflow), which were similarly ground-breaking in different ways.
I like your list better, mostly because of the inclusion of SNOBOL, which I never used, but was one of the first programming languages I read about as a young child after a book about it caught my attention at a public library book sale because of the funny name.
The only languages I was familiar with before this were BASIC, Logo, and a bit of 6502 assembly, though I had only used the latter by hand-assembly and calling it from BASIC following an example in the Atari BASIC manual[1].
Also, it's hard for me to imagine how anyone could make a list of ground-breaking programming languages that doesn't include Fortran and COBOL (or FLOW-MATIC as the source of many of its innovations).
They should be placed alongside each other, because Self OOP model is quite different from Smalltalk, including how the graphical programming experience feels like.
For those that never seen it, there are some old videos (taken from VHS) on the language site, https://selflanguage.org/
> I don’t understand why self is placed in the list instead of smalltalk.
The article explains that:
> Smalltalk inherited the notion of a value and its type from earlier languages, and implemented the idea of a class. All objects had a class that gave their type, and the class was used to construct objects of that type. Self disposed of the notion of class and worked solely with objects. As this is a purer form, I have chosen Self as the type specimen for this ur-language.
Yes, but I still don't understand that explanation. Clearly self is a descendant of Smalltalk, that purified a part; but still is a descendant. At least I understand the "ur-" as indicating linage, more about time as features. For me is still backwards.
Although it didn't call it that, Simula-67 was basically object-oriented and both preceded and inspired Smalltalk. But syntactically it looks much like other Algol-inspired langages so it doesn't look that interesting at first glance.
My favorite subject when studying CompSci (TU Delft) was called "Concepts of programming languages". We learned C, Scala (for functional) and Javascript (prototypes).
It made learning Elixir years later much easier.
We also had a course that basically summed up to programming agents to play Unreal Tournament in a language called GOAL which was based on Prolog.
For years I've wanted to use Prolog but could not figure out how. I ended up making a spellcheck to allow LLM's to iterate over and fix the dismal Papiamentu they generate.
The Unreal Tournament was the coolest thing I've ever seen. I think they shut it down the year after mine. (Now they have boring old regular AI like everyone else!)
I haven't found a good use for Prolog, though I haven't put much effort into it. I admit I was much more impressed by GOAL though, and I didn't realize until recently that you can replicate the whole thing in a more "ordinary" language (and that this gives many benefits). D'oh!
I took a similar class in college, and I'm also glad I did, even though the professor was kinda rubbish.
Even having the thinnest surface level understanding of the other ur-languages is so useful (and even more-so with assembly). I can't do anything useful with them, but it helps keep you from the "when all you have is a hammer, every problem looks like a nail" trap if you're at least aware of the existence of screwdrivers.
Same, our professor was okay, but he introduced a new system for the course that year, Weblab. I see it still exists and I have no doubt it works flawless now but every exam except the last ones was a mess.
And regarding the screwdriver, I fully agree. Especially with AI I can use the right tool for any job, despite me not being able to use it directly.
Reminds me a bit of Bruce Tate’s approach in 7 languages in 7 weeks, which is where I first encountered Erlang.
I think from a historical perspective, describing COBOL and Fortran as part of the ALGOL family is a stretch, but I suppose it’s a good reminder that all history is reductive.
There's also (besides Tate's sequel of 7 more languages), Dimitry Zinoviev's 7 Obscure Languahes in Seven Weeks. I liked it a lot, even if it hurt my feelings a bit to have my beloved Forth be one of the obscure languages (the others were APL, SNOBOL, Occam, Simula, Starset, and M4) -- I'm old and nerdy, but hadn't even heard of Occam and Starset.
Perhaps better IMHO is "Strange code" by Ronald Kneusel (NoStarch, 2022) [0], which I found more didactic and developed. Please note than I'm quite a fan of this author's other books [1].
I also think going back farther is a stretch. The first assembly languages were imperative, but what made Algol, Fortran, and Cobol interesting were functions and other features that allowed complex programming. Algol has the most descendants but Fortran was the first imperative programming language.
Does anybody know whether Fortran is older or younger than Algol? From Wikipedia, it looks like they were both developed around 1957. Was there any overlap in the design?
Rather COBOL is a living fossil? And today's Fortran is the FORTRAN family with horizontal gene transfer from the Algol lineage of programming languages.
> Fortran is one of the reasons OpenCL lost to CUDA, and now even AMD and Intel have finally Fortran support on their own stacks, not Khronos based.
FWIW, I loved using CUDA-Fortran. I think the ease of use of array variables maps very well with the way CUDA kernels work. It feels much more natural than in C++.
I mean, programming languages do not live; and they do not "die", per se, either. Just the usage may go down towards 0.
COBOL would then be close to extinction. I think it only has a few niche places in the USA and perhaps a very few more areas, but I don't think it will survive for many more decades to come, whereas I think C or python will be around in, say, three decades still.
> family with horizontal gene transfer
Well, you refer here to biology; viruses are the most famous for horizontal gene transfer, transposons and plasmids too. But I don't think these terms apply to software that well. Code does not magically "transfer" and work, often you have to adjust to a particular architecture - that was one key reason why C became so dynamic. In biology you basically just have DNA, if we ignore RNA viruses (but they all need a cell for their own propagation) 4 states per slot in dsDNA (A, T, C, G; here I exclude RNA, but RNA is in many ways just like DNA, see reverse transcriptase, also found in viruses). So you don't have to translate much at all; some organisms use different codons (mitochondrial DNA has a few different codon tables) but by and large what works in organism A, works in organism B too, if you just look to, say, wish to create a protein. That's why "genetic engineering" is so simple, in principle: it just works if you put genes into different organisms (again, some details may be different but e. g. UUU would could for phenylalanine in most organisms; UUU is the mRNA variant of course, in dsDNA it would be TTT). Also, there is little to no "planning" when horizontal gene transfer happens, whereas porting requires thinking by a human. I don't feel that analogy works well at all.
I don't know; COBOL pops up in banking all the time and Fortran lives on in a lot of specialised engineering applications.
Is there anything to be gained from rewriting it in Rust? Five years ago would there have been anything to be gained from rewriting it in Haskell? Ten years ago would there have been anything to be gained from rewriting it in Ruby? Fifteen years ago would there have been anything to gain from rewriting it in Clojure? Twenty years ago would there have been anything to gain from rewriting it in Java?
C++ has Algol roots, but I think the C++ template metaprogramming style is an ur-language of its own. You could draw some parallels with ML maybe, but they came at it from a different direction.
This article is full of gross mistakes. For example it claims that Caml is "Cambridge ML" which is ridiculously false. Fact check every sentence. Really sad.
Supposedly it was named "caml" because the author was a smoker who enjoyed camel cigarettes, hence the joke that [later implementations](https://en.wikipedia.org/wiki/Caml#History) were named "Caml Light" and "Caml Special Light"
I always enjoy these summaries. I took my bachelor of computer science in the early 1990s. It covered a language in most of these categories.
We didn't learn APL (Who is teaching the use of those custom keyboards to 100s of young students for one semester?)
The processing power of systems at the time made it clear which language classes were practically useful and usable for the time and which were not.
Prolog ran like a dog for even simple sets of logic.
We had the best internet access and pretty powerful desktop systems for the time.
I'm still curious why we didn't learn smalltalk. Could have been the difficulty of submitting and marking a system in a particular state rather than a file of code :)
That's amazing. I used to live under an architecture student (our building). His command of design history was great. His command of maths? Well, not so much.
Lots of us are having fun identifying our choice for missing family :)
One I might suggest is scripting languages, defined loosely by programming tools which dispatch high-level commands to act on data pipelines: sed, AWK, the sh family, Perl, PowerShell, Python and R as honorary members. In practice I might say SQL belongs here instead of under Prolog, but in theory of course SQL is like Prolog. Bourne shell might be the best representative, even if it's not the oldest.
AWK et al share characteristics from ALGOL and APL, but I feel they are very much their own thing. PowerShell is quite unique among modern languages.
I agree with "learn different classes of languages". OCaml was a language in which finally a function was a (mathematical) function. Mathematica thought me to look at expressions themselves as inputs. PostScript (with its reverse Polish notation going beyond simple arithmetics) rewired by brain.
At the same time, I don't agree with that it does not matter if one picks "Java, C#, C++, Python, or Ruby". If your goal is to do quick sort, then well, it does not.
If you want to use language for something (not only for its sake), then it makes a day and night difference. A person who wants to do 3D games and being shown Ruby or a person wanting to do exploratory data science and deep learning and being given Java are likely to get discouraged.
Even tho I probably never will get paid for writing Rust, I have zero regrets about learning it - it tough me really think about data ownership in programs.
I used to think being an engineer meant learning many languages. Turns out most of them solve the same problems — the returns diminish quickly. Depth across the stack (UX, CI/CD, model internals) compounds in a way that a sixth language never will.
I wonder if Occam is worth a mention? It doesn't feel like anything else here, and is playing with its hardware synthesis descendants on a FPGA is another "mind expanding" paradigm.
Most old-timers here are familiar with a Prolog-variant: make. Anyone who's struggled over a complex Makefile wishes they had a more sane declarative language!
> "What do I mean when I say fundamentals? If you have an array or list of items and you’re going to loop over it, that is the same in any imperative language."
Err
"Fortran: The world’s first programming language standard opened the door to modern computing"
Folks might find the following useful for studying PLs;
1) Advanced Programming Language Design by Raphael Finkel - A classic (late 90s) book comparing a whole smorgasbord of languages.
2) Design Concepts in Programming Languages by Franklyn Turbak et al. - A comprehensive (and big) book on PL design.
3) Concepts, Techniques and Models of Computer Programming by Peter Van Roy et al. - Shows how to organically add different programming paradigms to a simple core language.
For any other array language novices, I've experimented with K and J but had the best experience so far with BQN. It is a bit on the Lispy side like K but much better documented, and I thought the APL-esque symbolic alphabet was mnemonically helpful enough while reading code to justify learning an editor keyboard integration. (Plus it's fun.)
I’ve very slowly been trying to do the “99 problems” list in each of these languages groups. It’s been a fun experience seeing the differences. Though I think I would need a larger, less algorithmic, project to really see each group’s strengths. Especially for the OOP group.
One thing the article didn’t touch on was SmallTalk’s live visual environment. It’s not a normal source code / text language.
I’ve never worked with SmallTalk professionally, so I could be mistaken. I know you can write/compile SmallTalk source code. But my understanding is that this wasn’t the way it’s intended to be used. I am under the impression that the live images were how the language designers intended it to be used. And that the live images are a better representation of the OOP/message passing paradigm.
Wasn't the way it's intended to be used to do what ?
"Within each project, a set of changes you make to class descriptions is maintained. … Using a browser view of this set of changes, you can find out what you have been doing. Also, you can use the set of changes to create an external file containing descriptions of the modifications you have made to the system so that you can share your work with other users."
1984 "Smalltalk-80 The Interactive Programming Environment" page 46
~
"At the outset of a project involving two or more programmers: Do assign a member of the team to be the version manager. … The responsibilities of the version manager consist of collecting and cataloging code files submitted by all members of the team, periodically building a new system image incorporating all submitted code files, and releasing the image for use by the team. The version manager stores the current release and all code files for that release in a central place, allowing team members read access, and disallowing write access for anyone except the version manager."
1984 "Smalltalk-80 The Interactive Programming Environment" page 500
- Forth: you can use PFE,Gforth for ANS Forth requeriments.
Or EForth if you reached high skills levels
where the missing stuff can be just reimplemented.
EForth under Muxleq: https://github.com/howerj/muxleq
I can provide a working config where a 90% of it
would be valid across SF.
If you follow the instructions on compiling s9, mlite it's similar
with MinC for Windows. If you are a Unix/Linux/Mac user, you already
know how to do that.
You got the whole docs in the TGZ file, and the web.
Altough Scheme and CL are different paths. CL's loop it's really, really complex and Scheme it's pretty much straightforward to understand. Any advanced CL user will have to implement Scheme's syntax (and an interpreter) as an exercise for PAIP. CL in CL... well, CL is too huge, T3X tried with Kilo Lisp 23 http://t3x.org/klisp/22/index.html and I doubt if anyone can even complete anything but the few starting chapters from Intro to Common Lisp with it.
Or for Lisp you might as well start with Emacs Lisp - you are going to use it for a decent environment unless you have the Common Lisp IDEs which you have to pay for or Racket.
Huge tip: if you use MCCLIM, install Ultralisp first and (ql-quickload 'mcclim)
later: it will give you a big speed boost. Big, not as the ones from Phoronix.
Actually big. From 'I can almost see redrawing on a really old ass netbook' to
'snappy as TCL/Tk' under SBCL.
As you can see, you don't need to pay thousands of dollars.
For Scheme, S9 just targets R4RS but as a start it's more than enough, and for SICP
you can install Emacs+Geiser+chicken Scheme and from any Linux/BSD: distro
command prompt, you run:
I would add another to the list, which is languages where every expression yields zero or more values, particularly `jq`. there are some antecedents in Icon and xquery, but these generally require explicitly opting into either production or consumption of value streams, where jq does this stream processing automatically from the ground up. (icon requires use of a suspend and needs an every clause to walk the generated values, xquery requires explicit 'for' statements over streams as many builtin operators fail on value streams)
in jq, the comma separates expressions, which independently yield values. a span of such expressions is called a 'filter', since they are always run by passing values from the prior filter into them (with the initial values sourcing from json objects on stdin, or an implicit null if you pass -n to the program).
binding variables in the language is similarly done for each value their source yields
$ jq -nc ' (1,2,3) as $A | $A + $A '
2
4
6
functions in the language are neat because you can choose to accept arguments as either early bound values, or as thunks, with the former prefixed with a $.
for example, this runs `. + 100` parameters context, with `.` as the 10,20,30 passed to it:
in the world there should only be one debate and one summary about programming languages, and everybody should read that as a starting point before they enter the fray.
for example, I would say what i was taught, that lisp is lists made out of cons cells, which contain references to either other lists/cons cells or to atoms. lisp doesn't "have" parentheses any more than lisp has the rest of the ASCII table, but parentheses were chosen as a convenient way to write list structure down in ascii form because parentheses nest in exactly the way lists do and lists are the important thing.
then there never needs to be another discussion of parentheses and how you feel about them except in a sub-discussion if anybody has any bright ideas about alternative ways to convert list structure to ascii on paper. you disliking parentheses does not mean you dislike lisp; it just means you don't understand lisp, like for instance you think lisp has something to do with ascii.
similar ideas apply to C, wherein the braces and semicolons do not matter, what matters is datatypes that correspond to von neumann architectures including especially pointers, and where binary representation is easily exposed (with a syntax that does not matter but for the record is &, |, ~, ^, <<, or >>. if you hate those choices for operator syntax you could easily convert your own C to use parenthetical lists of CamelCase names of LeftAddRight or whatever floats your boat instead of a + sign), just keep in the front of your mind how data is represented internally and how it's mapped to the syntax you like, and whatever list structure you choose to order the lines of code)
Ur-ness is not interesting, just as the city of Ur was no doubt a lot less interesting than the city of Rome. Study Algol only to learn about call-by-name, then forget it.
I could go on, but, like the ripples from a gold coin thrown in the bay, the effect will soon dissipate as if I had never written anything at all.
Behold, future children, this is the ur-comment that started the cognitive revolution in how we teach programming languages in the 32nd century. The text and its important idea were almost lost in the tides of history, but it reached us thanks to a few eccentric ur-linguists digging through the ruins of post-WWIV civilizational collapse.
Hmm, here's a thought... If you want to stand out, it doesn't matter that some things now are easier for everybody, what matters is that that you are able to get better results than others. Learning multiple languages gives you more the ability to use them. It improves your thinking, makes you a better coder, and more able to understand different techniques. LLMs are tools, to use them better than the next person you need to understand what to ask, and what a good answer looks like.
Exactly. If nothing else, writing a solver in Python or Java might take dozens or hundreds of lines more code than Prolog, so simply knowing what tools are best for what jobs helps you be a better developer, whether you're using a compiler or an agent.
Sure, but that's about thinking and describing in more high-level English, not individual programming languages. That era is over.
Advice to juniors (say) to spend time learning multiple programming languages over good command of a single one, deep expertise in LLM use and basic software engineering principles is going to severely undermine their value in an already tough field for entrants. For seniors there will generally already be a reasonable grounding in multiple paradigms; delving much further into legacy manual coding styles is going to see them leapfrogged by experts in modern (ie AI-assisted) approaches.
Not at all. That’s like saying learning how different kinds of engines work is redundant in the age of taxis. You don’t have to know any of this stuff in order to get from A to B. But if you want to understand the processes involved in getting there, or you maybe want to be the one that builds a better self-driving vehicle, this is where you should start.
(GP here) Its true that we need to be cautious about continuing to exercise our thinking muscles, undoubtedly. Ofc you can do that without using legacy techniques, but each to their own.
Agree (redundant, not obsolete), but there are better tools for the job in terms of the production value gained in terms time and mental energy spent in mastering them. You can certainly think less as tech becomes more powerful in a domain; I wouldn't advise that either.
We got to build mini versions of the first 4 languages (imperative, lisp, ML, Smalltalk) in the PL course at tufts which is now published as a textbook [1]. There used to be a prolog part that sadly got cut.
[1]: https://www.cambridge.org/ir/universitypress/subjects/comput...
Maybe a version with the Prolog part could show up on the Internet Archive?
Here’s the accompanying code on github but we never got to that part in class: https://github.com/nrnrnr/build-prove-compare-student-code
One correction I'd make to the article's taxonomy: Ruby is an object oriented language not an Algol. Its inspiration is Smalltalk, and much of the standard library naming comes from that route (eg collect rather than map).
Ruby is object oriented from the ground up. Everything (and I do mean everything) is an object, and method call is conceived as passing messages to objects.
While Ruby is most often compared to Python (an Algol), they come from very different evolutionary routes, and have converged towards the same point in the ecosystem. I think of Ruby as a cuddly Alpaca compared to Python's spitting camel.
Since Python introduced new style classes, it also became a pure OOP language, even though it might not look like it at "Hello World" level, all primitive types have become objects as well.
I love to point this out to OOP haters,
I have found the definition of OOP to be fuzzy. For example, I don't see why having methods would make a data type object oriented. I associate OOP with factories, inheritance, using classes in places that might be functions otherwise, and similar abstractions.
Perhaps this is the counterfactual: I program in Python regularly, but don't program in an OOP style; I use dataclasses and enums as the basis, in a way similar to Rust, which by some definitions can't do OOP. So, if Rust can't do OOP (assumption) and I can write Python and Rust with equivalent structure (Assumption), does that mean Python isn't strictly OOP?
Rust can definitely do OOP, not only I could easily convert "Raytracing in one Weekend" from C++ into Rust, while keeping the same design, Microsoft has no issues adding a Rust projection for COM/WinRT, which are OOP ABIs.
However that is not the same as Python, which is 100% OOP, when using basic stuff like numbers, from Python language semantics those things are object, and there is the whole machinery in place even for basic stuff like addition.
And yes OOP is fuzzy from CS point of view there are multiple approaches and it doesn't get reduced to the way C++ and Java do it, just like FP and LP are fuzzy as well.
Some folks would swear if it isn't Haskell or Prolog, than it isn't FP or LP, when a CS book and programming language evolution will be more fuzzy.
> if Rust can't do OOP (assumption)
Rust handles basic OOP, but not all of the characteristics seen in C++ or Java:
https://doc.rust-lang.org/book/ch18-01-what-is-oo.html
If that's enough to make a language pure OOP, then Common Lisp is also a pure OOP languge:
If you're class of `42, you're probably a Lisp graybeard.
Common Lisp Object System and The Art of Metaobject Protocol.
Yes.
This is very cool, and I did not know this. Thank you!
I wonder if my formal university python training predated this change (~2010), or if the professors were themselves unaware of this.
They were unaware of it, or unwilling to talk about it, article from 2002, about changes introduced in 2001
https://gnosis.cx/publish/programming/metaclass_1.html
https://www.python.org/download/releases/2.2/descrintro/
> I love to point this out to OOP haters
That seems like a pretty lame gotcha--saying "Aha! The language you write in uses your hated paradigm under the hood" seems to invite the immediate response of "So? I don't use it."
It is more about those that proudly use Python because it isn't an OOP language, yep those do exist.
I think the choice to identify a specific ur-language as "Object oriented" throws people off since OO is just a style of programming in the same way that procedural is. I don't think it's useful to say that Python and C++ are both the same kind of language because they both have multiple inheritance, rather that's just an observable commonality, like noticing that both Delhi and Vegas are too hot. Yeah, but I don't think that's because they're the same kind of place...
Object-oriented programming is less of a syntax and more of a philosophy. While Erlang’s syntax belongs to the ML family, its creator, Joe Armstrong, argued that Erlang aligns more closely with the original OOP philosophy defined by the inventor of Smalltalk than Java does: 'The essence of object-oriented programming is messaging, not classes and inheritance.'
Yeah, but the thing about Vegas is that it's really more of a dry heat
Compared to Delhi? Ok. But I've had a soaking uncomfortable shirt every time I've been to Vegas, while in Phoenix it evaporates quickly.
Aren't camels a Perl thing?
> Aren't camels a Perl thing?
That's a deep cut. :-)
For anyone reading this, O'Reilly was once legendary for their cover-art mascots.
Kids those days. There's always one explaining the joke to others.
They're just doing their bit for today's lucky 10,000. Which, if you're unaware, is a reference to this XKCD comic: https://xkcd.com/1053/
Ruby keywords are not objects.
Both object oriented and an Algol.
I might add another class of languages: those intended to express proofs, via the Curry-Howard correspondence. Lean is a primary example here. This could be considered a subclass of functional languages but it might be different enough to warrant a separate class. In particular, the purpose of these programs is to be checked; execution is only secondary.
Theorem proving and complex types are like extensions on an otherwise ordinary language:
- Agda, Idris, etc. are functional languages extended with complex types
- Isabelle, Lean, etc. are functional languages extended with complex types and unreadable interactive proofs
- Dafny etc. are imperative languages extended with theorems and hints
- ACL2 is a LISP with theorems and hints
Related, typeclasses are effectively logic programming on an otherwise functional/imperative language (like traits in Rust, mentioned in https://rustc-dev-guide.rust-lang.org/traits/chalk.html).
> Agda, Idris, etc. are functional languages extended with complex types
I think they are not. No amount of type level extensions can turn a regular functional language like Haskell into something suitable for theorem proving. Adding dependent types to Haskell, for example, doesn't suffice. To build a theorem prover you need to take away some capability (namely, the ability to do general recursion - the base language must be total and can't be Turing complete), not add new capabilities. In Haskell everything can be "undefined" which means that you can prove everything (even things that are supposed to be false).
There's some ways you can recover Turing completeness in theorem provers. You can use effects like in F* (non-termination can be an effect). You can separate terms that can be used in proofs (those must be total) from terms that can only be used in computations (those can be Turing complete), like in Lean. But still, you need the base terms to be total, your logic is done in the fragment that isn't Turing complete, everything else depends on it.
This is just wrong, you're being too didactic. Idris specifically lets you implement nontotal functions in the same way that Rust lets you write memory-unsafe code. The idea is you isolate it to the part of the program with effects (including the main loop), which the compiler can't verify anyway, and leave the total formally verified stuff to the business logic. Anything that's marked as total is proven safe, so you only need to worry about a few ugly bits; just like unsafe Rust.
Idris absolutely is a general-purpose functional language in the ML family. It is Haskell, but boosted with dependent types.
Random passerby chiming in: so this means you can write "regular" software with this stuff?
While reading TFA I thought the theorem stuff deserved its own category, but I guess it's a specialization within an ur-family (several), rather than its own family?
It definitely sounds like it deserves its own category of programming language, though. The same way Lojban has ancestry in many natural languages but is very much doing its own thing.
Yes Idris was meant to write regular code. F* is also meant to write regular code
But I think that the theorem prover that excels most at regular code is actually Lean. The reason I think that is because Lean has a growing community, or at least is growing much faster than other similar languages, and for regular code you really need a healthy ecosystem of libraries and stuff.
Anyway here an article about Lean as a general purpose language https://kirancodes.me/posts/log-ocaml-to-lean.html
> if you're a developer who wants to exploit the multiplicative factor of a truly flexible and extensible programming language with state of the art features from the cutting-edge of PL research, then maybe give Lean a whirl!
Does not sound that appealing to me. Sounds like little consistency and having to learn a new language for every project.
I mentioned this later
> You can separate terms that can be used in proofs (those must be total) from terms that can only be used in computations (those can be Turing complete), like in Lean
What I meant is that the part of Idris that lets people prove theorems is the non-total part
But, I think you are right. Haskell could go there by adding a keyword to mark total functions, rather than marking nontotal functions like Idris does. It's otherwise very similar languages
Haskell has liquid haskell to do that.
> To build a theorem prover you need to take away some capability (namely, the ability to do general recursion - the base language must be total and can't be Turing complete), not add new capabilities. In Haskell everything can be "undefined" which means that you can prove everything (even things that are supposed to be false).
Despite what the fanatical constructivists (as opposed to the ones who simply think it's pragmatically nice) seem to want us to think, it turns out that you can prove interesting things with LEM (AKA call/cc) and classical logic.
How does this disagree with "the base language must be total and can't be Turing complete"
I mean you are kinda right but kinda wrong. To get a proof checker you take a typed lambda calculus and extend it with Pi, Sigma, and Indexed Inductive types while maintaining soundness.
Yes haskell's `bottom` breaks soundness, but that doesn't mean you need to take away some capability from the language. You just need to extend the language with a totality checker for the new proof fragment.
Lean is definitely a dependently typed ML-family language like Agda and Idris, so "ML" has it covered. And the long-term goal of Lean certainly is not "execution is only secondary"; Microsoft is clearly interested in writing real software with it: https://lean-lang.org/functional_programming_in_lean/Program...
OTOH if you really want to emphasize "intended to express proofs" then surely Prolog has that covered, so Lean can be seen as half ML, half Prolog. From this view, the Curry-Howard correspondence is just an implementation detail about choosing a particular computational approach to logic.
These are not true programming languages because by definition they are not Turing complete. If they were Turing complete it would be possible to write a false proof that just compiled down to a non-terminating program.
There has never been a requirement for a "true" programming language to be Turing complete.
Also, basically every such language has escape hatches similar to unsafe in Rust to allow expressions that are not provably terminating.
They can then just be accepted as an axiom.
I feel like Turing completeness has always been set as the boundary of programming language if there's any boundary at all. That's what people has been using to not include HTML as programming language for example. Or to include MTG as one.
I think it's a pretty recent thing. Turing completeness is neither a sufficient nor a required property.
HTML is not a programming language, it's a markup language - it's in the name, and it's the way it is used (it's not used to describe any kind of computation, it's straight up data that is parsed by algorithms).
Neither is PowerPoint, or game of life a programming language even though both are Turing complete.
Is Ladder Logic a programming language? It's used to program industrial controllers, and I don't believe it's Turing complete.
https://en.wikipedia.org/wiki/Ladder_logic
Turing completeness is the upper bound of computability, not the lower bound. It's useful mostly for showing that some thing can express the full range of computable problems, or for snarking that some thing is far more complex than it has any right to be.
Total languages omit partiality and non-termination from Turing completeness.
Partiality is IMO irrelevant when it comes to computability. Any partial function (that is, one whose range is not defined over its whole domain) can be expressed as a total function by either constricting the domain or expanding the range. For example, a "pop" operation on a stack is not defined for an empty stack. You can just loop forever if pop() is called on an empty stack. Alternatively, you can require that pop() is given a witness that the stack is non-empty, or you can require that pop() returns either the top-most element of the stack or a value that indicates the stack was empty. Both let you compute the same set of things as the former.
Non-termination is required to be Turing complete, because being Turing complete means being able to compute functions that one cannot reasonably expect to complete before the heat death of the universe. In _practice_ every function terminates when the computing process dies due to some external factor: process runs out of memory ("real" Turing machines have infinite memory!), user runs out of patience, machine runs out of power, universe runs out of stars, that sort of thing, so _in practice_ doing 2^64 iterations before giving up will generally* give you the same outcome as doing an unbounded number of iterations: it'll either terminate, or the process will be killed (here, due to reaching its iteration limit).
On the flip side, giving up non-termination and partiality only gives you increased correctness. If there's one thing we've definitely established in computing, it's that we will readily discard correctness to gain a little extra productivity. Why make a developer implement code to handle reaching an iteration limit when you can just make the user get sick of waiting and kill your app?
* 18 quintillion is a very large number. Have a try. The most trivial recursive function, on my M4 Mac, when convincing clang to be smart enough to turn it into a loop but dumb enough not to elide it altogether, would take a bit shy of 600 years to complete if iterating ULONG_MAX times; I didn't wait for that, if I'm honest with you, I ran it with a much smaller iteration count and multiplied it out.
This is all of them, properly speaking.
Incidentally, this is pretty much what Algol 60 was designed for and why to this day many academic papers use it or a closely related pseudocode.
These fall directly out of ML.
I recently revisited a language comparison project, a specific benchmark tallying the cycle decompositions in parallel of the 3,715,891,200 signed permutations on 10 letters. I kept a dozen languages as finalists, different philosophies but all choices I could imagine making for my research programming. Rather than "ur" I was looking for best modern realizations of various paradigms. And while I measured performance I also considered ease of AI help, and my willingness to review and think in the code. I worked hard to optimize each language, a form of tourism made possible by AI.
The results surprised me:
https://github.com/Syzygies/CompareNaively this is quite surprising, but the devil is in the details. With the exception of Lean I'd point out they're all fairly close: Chez being 2.5x slower than C++ is not ignorable but it's also quite good for a dynamically-typed JITted language[1]. And I'm not surprised that F# does so well at this particular task. Without looking into it more closely, this seems to be a story about F# on .NET Core having the most mature and painless out-of-the-box parallelism of these languages. I assume this is elapsed time, it would be interesting to see a breakdown of CPU time.
I don't think these results are quite comparable because of slightly differing parallelism strategies; I'd expect the F# implementation of just spinning off threads to be more a little more performant than a Rayon parallel iterator, which presumably has some overhead. But that really just shows how hard it is to do a cross-language comparison; Rust and C++ can certainly be made faster than the F# code by carefully manipulating a ton of low-level OS concurrency primitives. This would arguably also be little misleading. Likewise Chez and Haskell have good C FFI; does that count? It's a tricky and highly qualitative analysis.
[1] FYI, one possible performance improvement with the Chez code is keeping the permutations in fxvectors and replace math operations with the fixnum-specific equivalent - this tells the compiler/interpreter that the data are guaranteed to be machine integers rather than bigints, so they aren't boxed/unboxed. I am not sure without running it myself, but there seems to be avoidable allocations in the Chez implementation. https://cisco.github.io/ChezScheme/csug/objects.html#./objec...
Thank you. I will try your Chez idea. I love Chez, even if coding in Scheme can feel like rubbing sticks together to start a fire on an island, when e.g. Scala has induction ranges. And I didn't try Idris or Racket as they compile to Chez, but perhaps they do so better than I did.
As for parallelism this is a primary concern of mine, and I tried multiple approaches for every language where there was a choice. I used my own work-stealing code only when it beat standard libraries. AI warned me I was in over my head, that writing such a library takes years of experience, but my use case (and my expected use cases in my research) is so uniform that simple can win, minimally touching the required bases such as permuting tasks to avoid false sharing.
I don't believe that the JIT languages (F# on top) do so well because of better parallelism. This is branch optimization. For this use case an AOT compiler with ample benchmark data to influence output should do better. That isn't a thing, and the argument seems to be that few use cases stay consistent. A JIT can adapt.
Yes, Chez improved a bit, at the expensive of readability.
Yeah :/ For a larger program you can pay the readability toll once, via a syntactic form that expands the general vector/arithmetic operations to the fixnum versions, e.g. used something like
Sorry if I borked the indentation. I have been working on stuff like this, and more general macros around dependency injection and inversion of control (e.g. you could write this macro to take the type as a parameter and generate code optimized for 'bigint or 'rational). Maybe check back after the summer :)And BTW I misspoke earlier, of course Chez is AOT rather than JIT. From one approach it's sort of a hybrid: really fast on-the-fly AOT kinda looks like JIT, tongue-in-cheek you could say "NoT compilation" (nick-of-time). But proper JIT of course has huge advantages. If you reeaaaallly wanted to sabotage readability, Chez makes it easy to invoke the compiler at runtime, so along with the C FFI I think you could hack together some sort of JIT. But wow, what a mess that would be! You'd better be getting a PhD thesis out of it :) And if the performance is that critical you'd be much better off with F#.
I haven't looked into the code, but Lean being so slow may be misleading depending on how you benchmarked it. IMO the fairest test is how "Lean code" (or Rocq code, etc.) is actually run, which is as native C code following extraction.
Given the sane C defaults that are applied by code extraction techniques, the delta really shouldn't be so great. But it's a common pitfall to torture one's own verified code in order to get it proven, and I'm also not sure how good of support there is for parallelism.
None of the other comments are about program benchmarks.
The common theme is diverse notions of a language. When I order from a menu, I don't order based on price, but I prefer to see the prices.
Lean 4 is the most interesting language on my list. I didn't reject it on price.
there's a few more semantic families: verilog, petri nets and variants, Kahn process networks and dataflow machines, process calculi, reactive, term rewriting, constraint solvers/theorem provers (not the same with Prolog), probabilistic programming,
plus up and coming (actual production-ready) languages that don't fit perfectly in the 7 categories: unison, darklang, temporal dataflow, DBSP
It may feel like a little bit of cheating mentioning the above ones, as most are parallel to the regular von Neumann machine setup, but was meaning for a while to do an article with 'all ways we know how to compute (beyond von Neumann)'.
> was meaning for a while to do an article with 'all ways we know how to compute (beyond von Neumann)'.
Would be very glad to read this.
In the meantime, I reproduce a part of an article by Steve Yegge:
---
What Computers Really Are
Another realization I had while reading the book is that just about every course I took in my CS degree was either invented by Johnny von Neumann, or it's building on his work in mostly unintelligent ways.
Where to start? Before von Neumann, the only electronic computing devices were calculators. He invented the modern computer, effectively simulating a Universal Turing Machine because he felt a sequential device would be cheaper and faster to manufacture than a parallel one. I'd say at least 80% of what we learned in our undergrad machine-architecture course was straight out of his first report on designing a programmable computer. It really hasn't changed much.
He created a sequential-instruction device with a fast calculation unit but limited memory and slow data transfer (known as the infamous "von Neumann bottleneck", as if he's somehow responsible for everyone else being too stupid in the past 60 years to come up with something better. In fact, Johnny was well on his way to coming up with a working parallel computer based on neuron-like cellular automata; he probably would have had one in production by 1965 if he hadn't tragically died of cancer in 1957, at age 54.)
Von Neumann knew well the limitations of his sequential computer, but needed to solve real problems with it, so he invented everything you'd need to do so: encoding machine instructions as numbers, fixed-point arithmetic, conditional branching, iteration and program flow control, subroutines, debugging and error checking (both hardware and software), algorithms for converting binary to decimal and back, and mathematical and logical systems for modelling problems so they could be solved (or approximated) on his computing machine.
-Steve Yegge, Math Every Day
https://archive.ph/6tOQF
Von Neumann may possibly have been the smartest man to ever live, but giving him credit for all of this is too much, brushing aside many other inventors (oft independent, to his credit).
>term rewriting
In uni we had to make a spreadsheet software.
I volunteered to do the formula parser, thinking it sounded like a fun challenge.
I was stumped for a week, until I realized I could rewrite the formulas into a form I knew how to parse. So it would rewrite 1+1 into ADD(1,1) and so on.
I also refused to learn regex, so the parsing code was "interesting" ;)
I recall a comment from a colleague. "Okay, Andy says it works. Don't touch it." XD
Guy from another group used regex and his solution was 20x shorter than mine.
Regular expressions are probably not enough for parsing formulas (depending of course on the exact task given), they usually are at least a context free language.
Regular expressions are definitely enough for turning characters into tokens, after which a simple recursive descent parser is vastly more straightforward to write. Lexing is optional, but generally advised.
> plus up and coming (actual production-ready)
Plus up and coming (not quite production-ready IMO, but used in production anyways): ChatGPT and the like.
Of course, it’s debatable whether they are programming languages, but why wouldn’t they be. They aren’t deterministic, but I don’t think that is a must for a programming language, and they are used to let humans tell computers what to do.
Logic programming in S9 scheme:
https://www.t3x.org/amk/index.html
You can just get the code without buying the book, learn with Simply Scheme or any other book and apply the functions from the code, the solvers are really easy to understand.
also Sussman's propagators are nice to check out [0]
[0] The Art of the Propagator (mit url down for the moment)
Great list of languages that don't fit the conventional families. I've been curious about some of them, like Petri nets and term rewriting, and will enjoy exploring the others.
Found a working link to the paper about propagators.
The Art of the Propagator, Alexey Radul and Gerald Jay Sussman. https://groups.csail.mit.edu/mac/users/gjs/6.945/readings/ar... (PDF)
pure [https://agraef.github.io/pure-lang/] is probably the most "practical" term rewriting language, though mathematica is the most used one by far.
<snark>So they reinvented speadsheets?</snark>
I wrote something similar here: https://fmjlang.co.uk/blog/GroundBreakingLanguages.html
We agree on Algol, Lisp, Forth, APL, and Prolog. For ground-breaking functional language, I have SASL (St Andrews Static Language), which (just) predates ML, and for object oriented language, I have Smalltalk (which predates Self).
I also include Fortran, COBOL, SNOBOL (string processing), and Prograph (visual dataflow), which were similarly ground-breaking in different ways.
I like your list better, mostly because of the inclusion of SNOBOL, which I never used, but was one of the first programming languages I read about as a young child after a book about it caught my attention at a public library book sale because of the funny name.
The only languages I was familiar with before this were BASIC, Logo, and a bit of 6502 assembly, though I had only used the latter by hand-assembly and calling it from BASIC following an example in the Atari BASIC manual[1].
Also, it's hard for me to imagine how anyone could make a list of ground-breaking programming languages that doesn't include Fortran and COBOL (or FLOW-MATIC as the source of many of its innovations).
[1] https://archive.org/details/atari-basic-reference-manual/pag...
I don’t understand why self is placed in the list instead of smalltalk. Smalltalk came first, and Alan Key was the one who invented the “OOP” name.
Also ML is seen as a child of Lisp.
They should be placed alongside each other, because Self OOP model is quite different from Smalltalk, including how the graphical programming experience feels like.
For those that never seen it, there are some old videos (taken from VHS) on the language site, https://selflanguage.org/
> I don’t understand why self is placed in the list instead of smalltalk.
The article explains that:
> Smalltalk inherited the notion of a value and its type from earlier languages, and implemented the idea of a class. All objects had a class that gave their type, and the class was used to construct objects of that type. Self disposed of the notion of class and worked solely with objects. As this is a purer form, I have chosen Self as the type specimen for this ur-language.
Yes, but I still don't understand that explanation. Clearly self is a descendant of Smalltalk, that purified a part; but still is a descendant. At least I understand the "ur-" as indicating linage, more about time as features. For me is still backwards.
Although it didn't call it that, Simula-67 was basically object-oriented and both preceded and inspired Smalltalk. But syntactically it looks much like other Algol-inspired langages so it doesn't look that interesting at first glance.
My favorite subject when studying CompSci (TU Delft) was called "Concepts of programming languages". We learned C, Scala (for functional) and Javascript (prototypes).
It made learning Elixir years later much easier.
We also had a course that basically summed up to programming agents to play Unreal Tournament in a language called GOAL which was based on Prolog.
For years I've wanted to use Prolog but could not figure out how. I ended up making a spellcheck to allow LLM's to iterate over and fix the dismal Papiamentu they generate.
I was there, too. o/
The Unreal Tournament was the coolest thing I've ever seen. I think they shut it down the year after mine. (Now they have boring old regular AI like everyone else!)
I haven't found a good use for Prolog, though I haven't put much effort into it. I admit I was much more impressed by GOAL though, and I didn't realize until recently that you can replicate the whole thing in a more "ordinary" language (and that this gives many benefits). D'oh!
Hi there! Which year was that? I followed the course in 2012/2013. Bummer it's gone.
We almost won the tournament, we lost cause we overestimated the enemy.
We programmed our agents to assume that if they have our flag they're bringing it to their base, thus we sent all agents there to await their arrival.
And so they waited while our opponents ran in circles with our flag at the center or the map.
I tried many things in Prolog but ordinary languages often proved to be more suitable.
I recently vibe coded spellcheck.boneiru.online which is fully based on Prolog.
I realized a spellchecker is a perfect use case, since I basically need to check ortography which is a set of facts.
In terms of GOAL the text input would be the perception, and I then resolved whether the goal (correct text) is achievable.
The facts are all valid words in the language and the rules I got from an ortography book.
I took a similar class in college, and I'm also glad I did, even though the professor was kinda rubbish.
Even having the thinnest surface level understanding of the other ur-languages is so useful (and even more-so with assembly). I can't do anything useful with them, but it helps keep you from the "when all you have is a hammer, every problem looks like a nail" trap if you're at least aware of the existence of screwdrivers.
Same, our professor was okay, but he introduced a new system for the course that year, Weblab. I see it still exists and I have no doubt it works flawless now but every exam except the last ones was a mess.
And regarding the screwdriver, I fully agree. Especially with AI I can use the right tool for any job, despite me not being able to use it directly.
This GOAL?
https://en.wikipedia.org/wiki/Game_Oriented_Assembly_Lisp
No this GOAL: https://en.wikipedia.org/wiki/GOAL_agent_programming_languag...
Reminds me a bit of Bruce Tate’s approach in 7 languages in 7 weeks, which is where I first encountered Erlang.
I think from a historical perspective, describing COBOL and Fortran as part of the ALGOL family is a stretch, but I suppose it’s a good reminder that all history is reductive.
There's also (besides Tate's sequel of 7 more languages), Dimitry Zinoviev's 7 Obscure Languahes in Seven Weeks. I liked it a lot, even if it hurt my feelings a bit to have my beloved Forth be one of the obscure languages (the others were APL, SNOBOL, Occam, Simula, Starset, and M4) -- I'm old and nerdy, but hadn't even heard of Occam and Starset.
Perhaps better IMHO is "Strange code" by Ronald Kneusel (NoStarch, 2022) [0], which I found more didactic and developed. Please note than I'm quite a fan of this author's other books [1].
[0] https://nostarch.com/strange-code
[1] https://nostarch.com/search/Kneusel
I also think going back farther is a stretch. The first assembly languages were imperative, but what made Algol, Fortran, and Cobol interesting were functions and other features that allowed complex programming. Algol has the most descendants but Fortran was the first imperative programming language.
Does anybody know whether Fortran is older or younger than Algol? From Wikipedia, it looks like they were both developed around 1957. Was there any overlap in the design?
Algol was published in 1958, and FORTRAN in 1957. I think it's fair to say they were developed concurrently.
Rather COBOL is a living fossil? And today's Fortran is the FORTRAN family with horizontal gene transfer from the Algol lineage of programming languages.
Both languages have their standards updated still, latest year in both cases was 2023.
Fortran is one of the reasons OpenCL lost to CUDA, and now even AMD and Intel have finally Fortran support on their own stacks, not Khronos based.
https://developer.nvidia.com/cuda-fortran
Whereas Cobol, even has cloud and microservices.
https://www.rocketsoftware.com/en-us/products/cobol/visual-c...
https://aws.amazon.com/mainframe/
Incredible how being monetary relevant keeps some languages going.
Also note how the beloved UNIX and C are from 1971 - 73, only about 10 younger than COBOL.
> Fortran is one of the reasons OpenCL lost to CUDA, and now even AMD and Intel have finally Fortran support on their own stacks, not Khronos based.
FWIW, I loved using CUDA-Fortran. I think the ease of use of array variables maps very well with the way CUDA kernels work. It feels much more natural than in C++.
Can COBOL be called a living fossil?
I mean, programming languages do not live; and they do not "die", per se, either. Just the usage may go down towards 0.
COBOL would then be close to extinction. I think it only has a few niche places in the USA and perhaps a very few more areas, but I don't think it will survive for many more decades to come, whereas I think C or python will be around in, say, three decades still.
> family with horizontal gene transfer
Well, you refer here to biology; viruses are the most famous for horizontal gene transfer, transposons and plasmids too. But I don't think these terms apply to software that well. Code does not magically "transfer" and work, often you have to adjust to a particular architecture - that was one key reason why C became so dynamic. In biology you basically just have DNA, if we ignore RNA viruses (but they all need a cell for their own propagation) 4 states per slot in dsDNA (A, T, C, G; here I exclude RNA, but RNA is in many ways just like DNA, see reverse transcriptase, also found in viruses). So you don't have to translate much at all; some organisms use different codons (mitochondrial DNA has a few different codon tables) but by and large what works in organism A, works in organism B too, if you just look to, say, wish to create a protein. That's why "genetic engineering" is so simple, in principle: it just works if you put genes into different organisms (again, some details may be different but e. g. UUU would could for phenylalanine in most organisms; UUU is the mRNA variant of course, in dsDNA it would be TTT). Also, there is little to no "planning" when horizontal gene transfer happens, whereas porting requires thinking by a human. I don't feel that analogy works well at all.
I don't know; COBOL pops up in banking all the time and Fortran lives on in a lot of specialised engineering applications.
Is there anything to be gained from rewriting it in Rust? Five years ago would there have been anything to be gained from rewriting it in Haskell? Ten years ago would there have been anything to be gained from rewriting it in Ruby? Fifteen years ago would there have been anything to gain from rewriting it in Clojure? Twenty years ago would there have been anything to gain from rewriting it in Java?
And so on, all the way back.
Analogies are inherently reductive; by deeply scrutinizing the differences you do not prove the analogy is useless.
C++ has Algol roots, but I think the C++ template metaprogramming style is an ur-language of its own. You could draw some parallels with ML maybe, but they came at it from a different direction.
This. Misses the compile-time evaluation boat completely, even though the proverbial "sufficiently smart compiler" is based on the idea.
This article is full of gross mistakes. For example it claims that Caml is "Cambridge ML" which is ridiculously false. Fact check every sentence. Really sad.
For those curious: Cambridge ML is a thing, but abbreviated CML[0]; and whilst Caml is part of the ML family, it appears to be unrelated to CML.
[0] https://www.cl.cam.ac.uk/teaching/1011/FoundsCS/usingml.html
Caml was originally an acronym for Categorical Abstract Machine Language [1].
[1] https://en.wikipedia.org/wiki/Caml
Supposedly it was named "caml" because the author was a smoker who enjoyed camel cigarettes, hence the joke that [later implementations](https://en.wikipedia.org/wiki/Caml#History) were named "Caml Light" and "Caml Special Light"
Previous discussion: https://news.ycombinator.com/item?id=35813496
May 4, 2023, to be clear. 323 comments.
Also 30 Sept 2021, 29 comments, <https://news.ycombinator.com/item?id=28704495>.
I would refer to the world _cognate_[0]. 'Fundamental programming cognates' sounds cool as a uni course.
[0] https://en.wiktionary.org/wiki/cognate
"Cognato/a" in Italian is brother/sister-in-law
I always enjoy these summaries. I took my bachelor of computer science in the early 1990s. It covered a language in most of these categories.
We didn't learn APL (Who is teaching the use of those custom keyboards to 100s of young students for one semester?)
The processing power of systems at the time made it clear which language classes were practically useful and usable for the time and which were not.
Prolog ran like a dog for even simple sets of logic.
We had the best internet access and pretty powerful desktop systems for the time.
I'm still curious why we didn't learn smalltalk. Could have been the difficulty of submitting and marking a system in a particular state rather than a file of code :)
> who
Yale :-) Alan Perlis' intro to CS at Yale back in the late 80s was an APL class (a relatively small one, though.)
A Smalltalk implementation provides:
So there are plenty of ways to submit code to be marked.See "OU LearningWorks: a customized programming environment for Smalltalk modules"
https://ieeexplore.ieee.org/document/841064
Virginia Tech at least used to - the school of Architecture had a programming in APL class.
That's amazing. I used to live under an architecture student (our building). His command of design history was great. His command of maths? Well, not so much.
Lots of us are having fun identifying our choice for missing family :)
One I might suggest is scripting languages, defined loosely by programming tools which dispatch high-level commands to act on data pipelines: sed, AWK, the sh family, Perl, PowerShell, Python and R as honorary members. In practice I might say SQL belongs here instead of under Prolog, but in theory of course SQL is like Prolog. Bourne shell might be the best representative, even if it's not the oldest.
AWK et al share characteristics from ALGOL and APL, but I feel they are very much their own thing. PowerShell is quite unique among modern languages.
I'd add dataflow "languages" such as Excel and LabVIEW.
I agree with "learn different classes of languages". OCaml was a language in which finally a function was a (mathematical) function. Mathematica thought me to look at expressions themselves as inputs. PostScript (with its reverse Polish notation going beyond simple arithmetics) rewired by brain.
At the same time, I don't agree with that it does not matter if one picks "Java, C#, C++, Python, or Ruby". If your goal is to do quick sort, then well, it does not.
If you want to use language for something (not only for its sake), then it makes a day and night difference. A person who wants to do 3D games and being shown Ruby or a person wanting to do exploratory data science and deep learning and being given Java are likely to get discouraged.
Even tho I probably never will get paid for writing Rust, I have zero regrets about learning it - it tough me really think about data ownership in programs.
I used to think being an engineer meant learning many languages. Turns out most of them solve the same problems — the returns diminish quickly. Depth across the stack (UX, CI/CD, model internals) compounds in a way that a sixth language never will.
I think that depends on the other five languages. Learning e.g. APL or Smalltalk if you are FORTRAN - C programmer would be very mind expanding.
Which category is TRAC? https://en.wikipedia.org/wiki/TRAC_(programming_language)
I wonder if Occam is worth a mention? It doesn't feel like anything else here, and is playing with its hardware synthesis descendants on a FPGA is another "mind expanding" paradigm.
Most old-timers here are familiar with a Prolog-variant: make. Anyone who's struggled over a complex Makefile wishes they had a more sane declarative language!
> "What do I mean when I say fundamentals? If you have an array or list of items and you’re going to loop over it, that is the same in any imperative language."
Err
"Fortran: The world’s first programming language standard opened the door to modern computing"
https://www.ibm.com/history/fortran
Folks might find the following useful for studying PLs;
1) Advanced Programming Language Design by Raphael Finkel - A classic (late 90s) book comparing a whole smorgasbord of languages.
2) Design Concepts in Programming Languages by Franklyn Turbak et al. - A comprehensive (and big) book on PL design.
3) Concepts, Techniques and Models of Computer Programming by Peter Van Roy et al. - Shows how to organically add different programming paradigms to a simple core language.
For any other array language novices, I've experimented with K and J but had the best experience so far with BQN. It is a bit on the Lispy side like K but much better documented, and I thought the APL-esque symbolic alphabet was mnemonically helpful enough while reading code to justify learning an editor keyboard integration. (Plus it's fun.)
I’ve very slowly been trying to do the “99 problems” list in each of these languages groups. It’s been a fun experience seeing the differences. Though I think I would need a larger, less algorithmic, project to really see each group’s strengths. Especially for the OOP group.
One thing the article didn’t touch on was SmallTalk’s live visual environment. It’s not a normal source code / text language.
> It’s not a normal source code / text language.
Do you think source code cannot be compiled and run from the command-line?
https://benchmarksgame-team.pages.debian.net/benchmarksgame/...
I’ve never worked with SmallTalk professionally, so I could be mistaken. I know you can write/compile SmallTalk source code. But my understanding is that this wasn’t the way it’s intended to be used. I am under the impression that the live images were how the language designers intended it to be used. And that the live images are a better representation of the OOP/message passing paradigm.
Wasn't the way it's intended to be used to do what ?
"Within each project, a set of changes you make to class descriptions is maintained. … Using a browser view of this set of changes, you can find out what you have been doing. Also, you can use the set of changes to create an external file containing descriptions of the modifications you have made to the system so that you can share your work with other users."
1984 "Smalltalk-80 The Interactive Programming Environment" page 46
"At the outset of a project involving two or more programmers: Do assign a member of the team to be the version manager. … The responsibilities of the version manager consist of collecting and cataloging code files submitted by all members of the team, periodically building a new system image incorporating all submitted code files, and releasing the image for use by the team. The version manager stores the current release and all code files for that release in a central place, allowing team members read access, and disallowing write access for anyone except the version manager."1984 "Smalltalk-80 The Interactive Programming Environment" page 500
That sounds fun! What are the 99 problems? I found language specific lists like https://wiki.haskell.org/H-99:_Ninety-Nine_Haskell_Problems Or is there a language agnostic list?
P-99: Ninety-Nine Prolog Problems by Werner Hett is the original. The site is apparenty no longer accessible, but here's a copy: https://www.ic.unicamp.br/~meidanis/courses/mc336/2009s2/pro...
It did my heart good to see Forth listed.
- Algol 68 docs: https://algol68-lang.org/resources 'a68g' it's a free as in freedom compiler.
- Forth: you can use PFE,Gforth for ANS Forth requeriments. Or EForth if you reached high skills levels where the missing stuff can be just reimplemented.
EForth under Muxleq: https://github.com/howerj/muxleq I can provide a working config where a 90% of it would be valid across SF.
Starting Forth, ANS version: https://www.forth.com/starting-forth/
Thinking Forth, do this after finishing SF: https://thinking-forth.sourceforge.net/
Also, Forth Scientific Library. You can make it working with both GForth and PFE, just read the docs.
Full pack: https://www.taygeta.com/fsl/library/Library.tgz
Helping Forth code for GForth/PFE. If you put it under scilib/fs-util.fs, load it with:
https://www.taygeta.com/fsl/library/fsl-util.fs- Lisp. s9fes, it will compile under any nix/Mac/BSD out there, even with MinC.
S9fes: http://www.t3x.org/s9fes/
Pick the bleeding edge version, it will compile just fine.
For Windows users: MinC, install both EXE under Windows. First, mincexe, then buildtools*exe: https://minc.commandlinerevolution.nl/english/home.html
Then get 7zip to decompress the s9fes TGZ file, cd to that directory, and run 'make'.
Run ./s9 to get the prompt, or ./s9 file.scm where file.scm it's the source code.
In order to learn Scheme, there's are two newbie recommended books before "SICP".
Pick any, CACS, SS, it doesn't matter, both will guide you before SICP, the 'big' book on Scheme:
Simply Scheme https://people.eecs.berkeley.edu/~bh/pdf/
Simply.scm file, select from ';;; simply.scm version 3.13 (8/11/98)' to '(strings-are-numbers #t)' and save it as simply.scm
https://people.eecs.berkeley.edu/~bh/ssch27/appendix-simply....
Concrete Abstractions
Book:
https://www.d.umn.edu/~tcolburn/cs1581/ConcreteAbstractions....
The SCM files needed to be (load "foo.scm") ed in the code in order to do the exercises:
https://github.com/freezoo/scheme-concabs
If you are en Emacs user, just read the Elisp intro, it will work for a different Lisp family but with similar design.
Spot the differences:
Scheme (like s9):
We try: Elisp/Common Lisp (as the web site shows): Same there: - Ok, ML like languages:https://www.t3x.org/mlite/index.html
If you follow the instructions on compiling s9, mlite it's similar with MinC for Windows. If you are a Unix/Linux/Mac user, you already know how to do that.
You got the whole docs in the TGZ file, and the web.
For Lisp one could also start with Common Lisp: A Gentle Introduction to Symbolic Computation (<https://www.cs.cmu.edu/~dst/LispBook/book.pdf>) and follow it with SBCL.
Yep, and after that I'd jump into PAIP, Paradigms of AI Programming.
Code: https://github.com/norvig/paip-lisp
The EPUB looks broken in my machine, try the PDF: https://commons.wikimedia.org/wiki/File:Peter_Norvig._Paradi...
Altough Scheme and CL are different paths. CL's loop it's really, really complex and Scheme it's pretty much straightforward to understand. Any advanced CL user will have to implement Scheme's syntax (and an interpreter) as an exercise for PAIP. CL in CL... well, CL is too huge, T3X tried with Kilo Lisp 23 http://t3x.org/klisp/22/index.html and I doubt if anyone can even complete anything but the few starting chapters from Intro to Common Lisp with it.
Or for Lisp you might as well start with Emacs Lisp - you are going to use it for a decent environment unless you have the Common Lisp IDEs which you have to pay for or Racket.
Eh, no. You have Elisp+cl-lib but SBCL too, and you can use Sly wth SBCL.
Of Lem with SBCL+Quicklisp:
https://lem-project.github.io/usage/common_lisp/
Huge tip: if you use MCCLIM, install Ultralisp first and (ql-quickload 'mcclim) later: it will give you a big speed boost. Big, not as the ones from Phoronix. Actually big. From 'I can almost see redrawing on a really old ass netbook' to 'snappy as TCL/Tk' under SBCL.
https://ultralisp.org/
As you can see, you don't need to pay thousands of dollars.
For Scheme, S9 just targets R4RS but as a start it's more than enough, and for SICP you can install Emacs+Geiser+chicken Scheme and from any Linux/BSD: distro command prompt, you run:
And, as a ~/.csirc file: To run SCM stuff for SICP: or Done. Get the SICP PDF and start doing SICP. You can use Emacs+Geiser with and read it from and do it everything from withing Emacs by running (pick chicken as the interpreter). Save your Emacs settings. Done.Reminds me of the six programming language memory models:
https://canonical.org/~kragen/memory-models/
Isn‘t FORTRAN also a ur-language? It was invented in 1957.
FORTRAN (1957), ALGOL (1958), and COBOL (1959) are similar.
Technically, FORTRAN is the oldest. One could argue that ALGOL is the most influential for language design.
Plankalkül
Another direction to explore logic languages is Datalog.
I think damalig falls cleanly into the prolog family.
I would add another to the list, which is languages where every expression yields zero or more values, particularly `jq`. there are some antecedents in Icon and xquery, but these generally require explicitly opting into either production or consumption of value streams, where jq does this stream processing automatically from the ground up. (icon requires use of a suspend and needs an every clause to walk the generated values, xquery requires explicit 'for' statements over streams as many builtin operators fail on value streams)
in jq, the comma separates expressions, which independently yield values. a span of such expressions is called a 'filter', since they are always run by passing values from the prior filter into them (with the initial values sourcing from json objects on stdin, or an implicit null if you pass -n to the program).
brackets collect values yielded inside of them. if you have a complex object that includes multiple expressions yielding multiple values, construction will permute over them. the pipe operator `|` runs the next filter with each value yielded by the prior, that value represented by the current value operator `.`. binding variables in the language is similarly done for each value their source yields functions in the language are neat because you can choose to accept arguments as either early bound values, or as thunks, with the former prefixed with a $.for example, this runs `. + 100` parameters context, with `.` as the 10,20,30 passed to it:
where this runs `. + 100` in the context of its use inside the function, instead receiving 1,2,3: so you could define map taking a current-value array and applying an expression to each entry like so: it's a fun little language for some quick data munging, but the semantics themselves are a decent reason to learn it.in the world there should only be one debate and one summary about programming languages, and everybody should read that as a starting point before they enter the fray.
for example, I would say what i was taught, that lisp is lists made out of cons cells, which contain references to either other lists/cons cells or to atoms. lisp doesn't "have" parentheses any more than lisp has the rest of the ASCII table, but parentheses were chosen as a convenient way to write list structure down in ascii form because parentheses nest in exactly the way lists do and lists are the important thing.
then there never needs to be another discussion of parentheses and how you feel about them except in a sub-discussion if anybody has any bright ideas about alternative ways to convert list structure to ascii on paper. you disliking parentheses does not mean you dislike lisp; it just means you don't understand lisp, like for instance you think lisp has something to do with ascii.
similar ideas apply to C, wherein the braces and semicolons do not matter, what matters is datatypes that correspond to von neumann architectures including especially pointers, and where binary representation is easily exposed (with a syntax that does not matter but for the record is &, |, ~, ^, <<, or >>. if you hate those choices for operator syntax you could easily convert your own C to use parenthetical lists of CamelCase names of LeftAddRight or whatever floats your boat instead of a + sign), just keep in the front of your mind how data is represented internally and how it's mapped to the syntax you like, and whatever list structure you choose to order the lines of code)
Ur-ness is not interesting, just as the city of Ur was no doubt a lot less interesting than the city of Rome. Study Algol only to learn about call-by-name, then forget it.
I could go on, but, like the ripples from a gold coin thrown in the bay, the effect will soon dissipate as if I had never written anything at all.
Behold, future children, this is the ur-comment that started the cognitive revolution in how we teach programming languages in the 32nd century. The text and its important idea were almost lost in the tides of history, but it reached us thanks to a few eccentric ur-linguists digging through the ruins of post-WWIV civilizational collapse.
you have not discovered the ur-comment, you have discovered the ruins of the pedagogy that was Ur. "look on my works, ye mighty, and despair"
I know where the gold coin is.
Rebol should be another family
[dead]
laugh in vibe coding
(2022) and unfortunately advice to spend significant amounts of time in learning multiple languages is becoming rapidly redundant in the LLM age.
Hmm, here's a thought... If you want to stand out, it doesn't matter that some things now are easier for everybody, what matters is that that you are able to get better results than others. Learning multiple languages gives you more the ability to use them. It improves your thinking, makes you a better coder, and more able to understand different techniques. LLMs are tools, to use them better than the next person you need to understand what to ask, and what a good answer looks like.
Exactly. If nothing else, writing a solver in Python or Java might take dozens or hundreds of lines more code than Prolog, so simply knowing what tools are best for what jobs helps you be a better developer, whether you're using a compiler or an agent.
Sure, but that's about thinking and describing in more high-level English, not individual programming languages. That era is over.
Advice to juniors (say) to spend time learning multiple programming languages over good command of a single one, deep expertise in LLM use and basic software engineering principles is going to severely undermine their value in an already tough field for entrants. For seniors there will generally already be a reasonable grounding in multiple paradigms; delving much further into legacy manual coding styles is going to see them leapfrogged by experts in modern (ie AI-assisted) approaches.
Not at all. That’s like saying learning how different kinds of engines work is redundant in the age of taxis. You don’t have to know any of this stuff in order to get from A to B. But if you want to understand the processes involved in getting there, or you maybe want to be the one that builds a better self-driving vehicle, this is where you should start.
These are tools for thinking with, so not obsolete.
GP is just trying to convince you that thinking is "rapidly redundant in the LLM age".
(GP here) Its true that we need to be cautious about continuing to exercise our thinking muscles, undoubtedly. Ofc you can do that without using legacy techniques, but each to their own.
Agree (redundant, not obsolete), but there are better tools for the job in terms of the production value gained in terms time and mental energy spent in mastering them. You can certainly think less as tech becomes more powerful in a domain; I wouldn't advise that either.
You're right. We should just run everything in java-script because that's what LLMs are good at right?
Make sure to let it pull in as many npm dependencies as it can.