09:09 <hanker> contificate: couldn't agree more, typescript was the industry's biggest mistake

11:38 <discocaml> <empathy.monster> This page links from the OCaml Learn section on the website: https://en.m.wikipedia.org/wiki/Expression-oriented_programming_language

11:40 <discocaml> <empathy.monster> Under the section Criticism and Prevention is this statement: In some expression-oriented languages, expressions that merely cause side effects return void types.

11:40 <discocaml> <empathy.monster>

11:40 <discocaml> <empathy.monster> I'm trying to understand, is this just merely a statement of fact, a criticism, or prevention? Or perhaps neither?

11:59 <discocaml> <edhebi> considering the number of notes on this page, I'm really not sure anyone knows

12:06 <discocaml> <empathy.monster> Yeah, wikipedia can be a mess sometimes

12:10 <discocaml> <empathy.monster> Okay, well here's what I'm trying to understand. If a language allows only an if (here's the question, statement or expression?) that's not required to be followed by an else block, and that if only prints to the console and returns void, is that an if statement or expression? Because that's not my understanding of an expression, but I was told it is an expression. And in the Ocaml Learn section, and even Haskell material, care is usua

12:12 <discocaml> <empathy.monster> The language is imperative, but it's expression-oriented

12:17 <discocaml> <biscotte9682> The `if` construct in ocaml is always an expression, `if b then e` without `else` is syntax sugar for `if b then e else ()`, however, if you add `;` at the end it becomes a statement

12:18 <discocaml> <empathy.monster> Hi, thank you. I do know that. I'm referring to an imperative language

12:19 <discocaml> <empathy.monster> Your thumbnail lol

12:19 <discocaml> <empathy.monster> I just watched that the other day actually

12:26 <discocaml> <empathy.monster> Let's take the above example: if b then e else () as an expression. Let's take the else clause. That evaluates to a value of unit, or type of unit? If it's type, I'm not understanding how 'evaluating' to a type would be considered an expression, bc my understanding is an expression evaluates to a value.

12:29 <discocaml> <empathy.monster> Or is unit both a type and a value at the same time?

12:30 <discocaml> <octachron> `()` is the one and only value of type `unit`.

12:30 <discocaml> <empathy.monster> Okayyyyy

12:31 <discocaml> <octachron> Thus people tends to name the value `()` unit too from time to time.

12:32 <discocaml> <octachron> because the separation between a category that contains one element and this one element becomes rather hazy.

12:33 <discocaml> <empathy.monster> I see, that clarifies is for me though, ty

12:33 <discocaml> <empathy.monster> At least for that ocaml expression. I'm still confused as to my original question

12:35 <discocaml> <octachron> Note that `void` is really ambiguous as a type name, because the vernacular definition may point to an empty type, whereas the C use corresponds rather to `unit` (aka a type with one inhabitant which thus contains zero information).

12:36 <discocaml> <empathy.monster> That makes sense

12:36 <discocaml> <edhebi> void is also highly irregular and special-cased all over the place in C and C++

12:37 <discocaml> <empathy.monster> I'll have to dig into the language spec it seems, or do some introspection on it

12:37 <discocaml> <edhebi> to the point that in a lot of language, `()` means both the type and the value

12:37 <discocaml> <empathy.monster> The language my question was about is called Hare

12:38 <discocaml> <octachron> The separation between statements and expressions is a matter of convention. In an expression-oriented language, the convention will often be that every value-level constructs are expressions. Thus `if true then ()` will be an expression.

12:39 <discocaml> <empathy.monster> So if void in Hare is more like unit, then it'd be an expression, is my understanding then. Although again, void is really ambiguous

12:40 <discocaml> <edhebi> hard to say much without known their language

12:40 <discocaml> <empathy.monster> Ah ha, okay

12:41 <discocaml> <empathy.monster> I think there's some type introspection stuff in the stdlib I'll play around with

12:41 <discocaml> <edhebi> a quick look at their doc seems to imply that they go the C route where void isn't considered a type in most places

12:43 <discocaml> <edhebi> > **The `void` type**

12:43 <discocaml> <edhebi> >

12:43 <discocaml> <edhebi> > 6.6.12.15 The voidtype represents an object with no storage. Only one value with this type exists (the `void` value); this value is also the default value.

12:43 <discocaml> <edhebi> >

12:43 <discocaml> <edhebi> > 6.6.12.16 The size and alignment of `void` shall be zero.

12:43 <discocaml> <edhebi> looks like a plain unit type

12:48 <discocaml> <empathy.monster> I did read that in the spec, but I'm still learning. You pointing that out is very helpful, bc unit in ocaml is very easy for me to grasp, so that comparison is perfect

12:48 <discocaml> <empathy.monster> I really appreciate it, everyone

14:13 <discocaml> <JM> thanks

14:19 <discocaml> <hungrystackofpancakes> hello!

14:19 <discocaml> <hungrystackofpancakes> i am here to learn ocaml

15:57 <discocaml> <contificate> Your Discord tag says "C++" so I feel it's my ethical duty to tell you that there's a genuine possibility you will dislike C++ after learning OCaml. 😼

15:58 <discocaml> <hungrystackofpancakes> after almost 4 years using c++

15:58 <discocaml> <hungrystackofpancakes> i tried like 20 languages

15:58 <discocaml> <hungrystackofpancakes> nothing replaced c++

15:58 <discocaml> <hungrystackofpancakes> lol

15:59 <discocaml> <contificate> do you write any symbolic computing software

15:59 <discocaml> <contificate> e.g. translators (or "compilers")

16:06 <discocaml> <hungrystackofpancakes> i use lisp

16:08 <discocaml> <contificate> for the same things as you use C++?

16:08 <discocaml> <contificate> Lisps are mostly DOA for me as I prefer static typing, so OCaml is a better replacement

16:26 <discocaml> <barconstruction> I saw a lisp guy post on Reddit recently about something called Coalton, an add on for common lisp that gives static typing features

16:27 <discocaml> <lost.ray> What you use C++ for?

16:27 <discocaml> <hungrystackofpancakes> everything

16:27 <discocaml> <lost.ray> For web frontend also?

16:28 <discocaml> <hungrystackofpancakes> only backend

16:28 <discocaml> <hungrystackofpancakes> frontend is ju9st html file

16:28 <discocaml> <contificate> do you ever think "damn, it'd be neat if I could economically specify and discern intuitively defined data"

16:28 <discocaml> <hungrystackofpancakes> frontend is just html file

16:28 <discocaml> <contificate> inductively* sorry on phone

16:29 <discocaml> <barconstruction> I think you could probably get away with seven lines of CSS too

16:30 <discocaml> <lost.ray> Maybe it's time to use WASM C++ and write some cool CSR app. Then you can do the same in Typescript with Vue/React/Svelte and compare DX. I think the choice will be obvious

16:31 <discocaml> <contificate> ergonomically* this phone is terrible

16:31 <discocaml> <barconstruction> No, economically works too. Economy, you know, parsimony with text

16:31 <discocaml> <contificate> economic in sense of time it takes C++ programmers to specify ASTs

16:32 <discocaml> <barconstruction> Yeah.

16:32 <discocaml> <contificate> a few hours of work to write out a few OCaml variant types

16:32 <discocaml> <contificate> real programming

16:33 <discocaml> <contificate> if you don't get carpal tunnel before you even start writing IR transforms, are you even programming C or C++

16:33 <discocaml> <polyml> with C++ you don't have to get carpal tunnel because compiling can rest your hands

16:34 <humasect> shift+ shift+ shift+

16:34 <humasect> in emacs i use snoopy-mode , so the number row (1234567890) already shifted

16:35 <humasect> (and also, anyway, i use the symbols much more often than the default of numbers.)

16:35 <humasect> who types literal constants unless they are filling in forms in the office ?

16:35 <discocaml> <lost.ray> Is it some kind of high load work? Or maybe very low latency, something like HFT?

16:36 <discocaml> <hungrystackofpancakes> it is the only lang that i actually know how to write well

16:36 <discocaml> <contificate> unrelated but I use capslock and I've never managed to hold shift to replace it

16:36 <humasect> ah my capslock is set to ctrl

16:36 <discocaml> <contificate> do you use evil mode

16:36 <humasect> ButWritingWordsLikeThis::can::getAnnoying::someTimes((()))

16:36 <humasect> nope vanilla

16:36 <discocaml> <contificate> yeah I like snake_case in C++

16:37 <discocaml> <barconstruction> I don't think any language is necessarily more suitable for web backend. It's not really convincing to me that javascript should be more usable for backend because it's the only language usable for frontend

16:37 <humasect> ocaml is snake case as well , but i co-erce it into camelStyle (ironical?)

16:37 <discocaml> <contificate> I don't think JS is suitable for any business logic beyond spewing garbage onto the DOM

16:37 <humasect> its like english. gets job done

16:38 <humasect> lingua franca literally meant French as the world's speaking of global importances

16:38 <discocaml> <lost.ray> Yeah, it's time to write SPA into C++

16:38 <discocaml> <lost.ray> Everyone must suffer

16:38 <humasect> i'm looking back at Objective-C this week, for Metal and to revive an old project (from memory) and i am sad that it is now called OCaml and not Objective-CAML =(

16:38 <discocaml> <lost.ray> Yeah, it's time to write SPA in C++

16:41 <humasect> SPA

16:42 <discocaml> <barconstruction> Ironic that the phrase "lingua Franca" is Latin. If French was so important then why isn't the phrase "lingua Franca" in French?

16:44 <humasect> because its already lingua-franca-ing if one speaking french .

16:44 <luc4> Hello! I'm trying to contribute to an open source project, but I'm getting this error: https://pastebin.com/rVe9s8rB. This happens when I triy to link to ocurl downloaded with opam. My understanding is that, for some reason, cdk.cmxa was built against a version of ocurl, while it is now trying to link to another. But I don't think I have two versions of ocurl in this system. Any other explaination for this error?

18:56 <discocaml> <jalsol> OCaml overtook C++ to become my favorite, but C++ is still a strong second to me lmao

18:59 <discocaml> <jalsol> as in it obviously could've been better, but there's nothing better than it (other than OCaml)

19:26 <discocaml> <contificate> depends what you're programming, as always - I went from loving to despising it in the space of a few weeks

19:50 <discocaml> <polyml> C++ is ok for consuming existing apis with dot/arrow syntax and autocompletion

19:50 <discocaml> <polyml> a lot of modern programming seems to be memorizing and consuming existing libraries instead of solving novel problems

19:52 <discocaml> <yawaramin> that's what the SICP people said to justify switching from Scheme to Python

19:59 <Anarchos> polyml indeed ! But the real next big step will be to write only specs in Coq/Lean/Agda... and let the spec be verified, and code autogenerated.

20:16 <discocaml> <jalsol> I've just landed a HFT intern, where C++ is probably the dominant language (yes, I really wish I could get into Jane Street)

20:18 <discocaml> <jalsol> there are things I don't like about C++ for sure, but the control over performance is a selling point for me

20:21 <discocaml> <jalsol> there is also metaprogramming, but it's a double-edged sword

20:24 <discocaml> <jalsol> I was not aware of the Python version; I know the JavaScript version exists (and I know one of the authors irl)

20:28 <discocaml> <jalsol> I was not aware of the Python version; I know the JavaScript version exists (and I know one of the authors irl, he literally taught JS as a modern Scheme with a custom JS compiler)

20:43 <discocaml> <yawaramin> i thought it would be cool to translate the 'Composing Programs' books from Python to OCaml, but they start _heavy_ –straight into making an HTTP request: https://www.composingprograms.com/pages/11-getting-started.html

21:07 <discocaml> <contificate> lol, C++ programmers go to lengths to avoid copying a string - unaware of what really dominates the performance of their programs

21:38 <companion_cube> Or you know, they profiled and it was costly

21:42 <discocaml> <contificate> nah I'm talking about the pedantry of the general community

21:44 <discocaml> <contificate> they teach C++ idioms using tables of info at cppcon

21:45 <discocaml> <empathy.monster> It's actually called rocq ("rock") now: https://rocq-prover.org/