02:58 <discocaml> <contextfreebeer> https://ocaml.org/papers

02:58 <discocaml> <contextfreebeer> Start here I guess, bound to find something

07:23 <discocaml> <jobhdez> Thanks

07:23 <discocaml> <jobhdez> What type of projects do you guys make with ocaml?

12:17 <type> hi! i am looking for a thorough description of the OCaml's garbage collector. i am aware of RWO's chapter and OCaml's wiki, but i was wondering if any of you could point me elsewhere.

12:17 <olle> type: source code?

12:17 <type> that would be my last resort

12:20 <type> in the sense that i am essentially trying to find "bugs" in the implementation

12:22 <Armael> could you say a bit more about what you're trying to do?

12:23 <discocaml> <contextfreebeer> the most notable OCaml projects to date have been compilers, proof assistants and SMT solvers, people that are interested in that kind of stuff tend to be drawn to OCaml, the biggest projects I've done personally are a toy compiler, SAT solver and a MIPS emulator (ocaml is not super great for writing emulators IME)

12:24 <discocaml> <contextfreebeer> but people make all kinds of things

12:25 <type> sure. i am basically trying to write formal specifications (temporal properties) of what the GC is supposed to do and check if it is doing the right thing based on its runtime behaviour.

12:26 <discocaml> <contextfreebeer> type: do you mean the new multithreaded GC?

12:27 <Armael> I think Damien Doligez was doing som of that in his PhD? (on top of the original author of the ocaml GC)

12:27 <type> yes, but i'm not focusing on multithreaded programs

12:27 <Armael> he seems a relevant person to talk to

12:28 <type> looks like a good start. thanks a lot for the pointer!

12:28 <Armael> on top of being the original author*

12:29 <Armael> np

12:34 <octachron> There is also Alexandre Moine's work (http://cambium.inria.fr/~amoine) on formal specification of languages with GCs

12:36 <type> many thanks! :)

12:39 <Armael> (there's also our formalization of (a subset of) the ocaml ffi in https://melocoton-project.github.io/ :p, which needs to account for the effect of the GC to some extent---thought I assume is related but not directly relevant)

12:42 <type> i will take a look at it. thanks!

14:12 <discocaml> <Ada> is ocaml getting more popular for web use?

14:12 <olle> Doubt it

14:12 <olle> Or, percentage or absolute numbers?

15:18 <discocaml> <bluddy5> It's difficult to say. Dream seems to be a fairly popular choice, and a solid library can attract people. ReasonML attracted many web developers, and after it imploded, many of those who remained seem to have switched to Melange, which is more closely tied to the OCaml ecosystem.

16:04 <discocaml> <deepspacejohn> to me it feels like it's getting more popular for web use in a technical sense, that more people are using it now then they were in the past. But that's completely anecdotal from my perspective. And it doesn't feel like it's close to competing with the other big web languages (yet?)

16:53 <discocaml> <noahtheduke> i didn't know reasonml imploded

17:14 <discocaml> <Ada> reasonml always seemed. like a bit of a bad idea to me (not to be rude)

19:17 <discocaml> <leostera> hi folks, anyone's got a snipper for to generate int32's from Mirage_crypto_rng? i can get a random Cstruct.t but, well, not what i want

19:17 <discocaml> <leostera> hi folks, anyone's got a snippet for/to generate int32's from Mirage_crypto_rng? i can get a random Cstruct.t but, well, not what i want

19:17 <discocaml> <leostera> hi folks, anyone's got a snippet to generate int32's from Mirage_crypto_rng? i can get a random Cstruct.t but, well, not what i want

19:20 <reynir> I think maybe random-conv is what you're looking for. One sec

19:23 <reynir> Sorry it was randomconv: `Randomconv.int32 Mirage_crypto_rng.generate`

19:23 <reynir> @leostera ^

19:25 <discocaml> <leostera> excellent, thanks @reynir!

20:20 <discocaml> <coollcat> I seriously doubt OCaml will ever compete with something like Java or ruby for web use, but that's okay tbh.

20:20 <discocaml> <coollcat> I find this language to be super opinionated - in a good way

20:38 <discocaml> <leostera> its really a matter of incentives

20:38 <discocaml> <leostera> there's tons of amazing stuff built in ocaml that just needs proper surfacing, focus on dx, docs – if the incentives aren't aligned to prioritize that kind of work, then it will never happen

20:41 <discocaml> <leostera> ruby in particular is a language that is _extremely_ opinionated – the language is so maleable, its too easy to find yourself writing your own dialect of it (or someone else's, *cough rails cough*)

20:44 <discocaml> <leostera> for ex. @reynir just helped me find a way to get a random int32 out of mirage_crypto, which is afaict _the_ crypto library for ocaml, and its surprisingly unintuitive for someone who doesn't want to care about Fortuna, or what Mirage is, or what a Cstruct is.

20:46 <discocaml> <leostera> so in many respects ocaml is a lot more about configurability than opinionated libs! (and this is great for a lot of use-cases)

21:30 <discocaml> <Ada> it is a multi paradigm language

22:31 <remexre> is there a way to see the code generated by PPX rewriters, for debugging purposes? (I'm using dune and haven't used ppx without it -- sorry if this is obvious!)

22:37 <discocaml> <Et7f3 (@me on reply)> I would run dune in verbose mode, copy the ppx invocation then see what options the ppx driver proposes or use ocamlopt invocation with -dparsetree

22:38 <discocaml> <Et7f3 (@me on reply)> (I haven't used ocaml in a while but it is a general method for everything)

22:47 <remexre> ah, cool, thanks

23:48 <discocaml> <deepspacejohn> `dune show pp` will also print the preprocessed output.