03:12 <discocaml> <dubious245> I do wish there was more ocaml youtube content.

03:27 <discocaml> <nothisisjohn> What sort of OCaml content are you (or anyone) interested in? I had a moderately successful channel a few years ago and have been toying with posting videos over the summer. I think that I would enjoy making / watching ~10 minute videos of implementing a small fun program. the first two that come to mind are 2048 and a wordle solver.

04:17 <discocaml> <dubious245> :bThoughtful:

04:18 <discocaml> <cod1r> janestrreet youtube is pretty good

04:18 <discocaml> <dubious245> I'd just like to see stuff being used. Not tutorials. Just here a thing this is how I solved it using ocaml.

05:10 <discocaml> <tornato> whoa

05:10 <discocaml> <tornato> my friend is #5 in the US on hackerrank

06:23 <discocaml> <.pseud.> Using stuff. Coming in, knowing a bit of the syntax and underlying theory. Getting started with ocaml is still quite difficult.

06:23 <discocaml> <.pseud.> Broken(?), severely undocumented libraries like caqti getting recommended is not helping.

06:23 <discocaml> <.pseud.> Dune and co is cool, but getting it all up and running is a bit hard, it is a bit alien to most.

06:23 <discocaml> <.pseud.> I would definitely spend time on getting going (dune, opam, maybe wrap in nix) then good editor plugins. Then maybe showcase other cool stuff like utop.

06:23 <discocaml> <.pseud.> From there, small, concise "building xyz in ocaml" 🙂

06:23 <discocaml> <.pseud.> CLI app, (arc parsing etc), maybe a telegram bot (http requests etc), database interaction. Something showing how to handle JSON and mapping into well defined types etc etc.

07:08 <discocaml> <regularspatula> Caqti is broken for you?

07:15 <discocaml> <.pseud.> Wrote in the beginner channel about it. 🙂

07:15 <discocaml> <.pseud.> Fwiw not alone, saw another blog post where caqti was also problematic for a person starting out with Dream .

07:15 <discocaml> <.pseud.> Yea. I would never recommend it, tbh.

07:18 <discocaml> <scope9247> functors define parametrized modules (compile-time), method calls happen via dynamic dispatch, so modules in this case for sure?

07:20 <discocaml> <regularspatula> Can you link the blog post?

07:31 <discocaml> <.pseud.> It's as much an issue of the API churn as anything. He just got caught out by one library using caqti needing the 2.x api while Dream at the time supported 1.x

07:31 <discocaml> <.pseud.> https://hypirion.com/musings/getting-ocaml-through-the-eye-of-a-needle

07:31 <discocaml> <.pseud.>

07:31 <discocaml> <.pseud.> My issue was that LLMs recommend what I assume was the earlier 1.x API. Not the library author's fault, I know.

07:31 <discocaml> <.pseud.>

07:31 <discocaml> <.pseud.> but then it's compounded by https://github.com/paurkedal/ocaml-caqti?tab=readme-ov-file#running-under-utop not actually working (you should `#require "caqti-lwt.unix"` - I saw mentions of `caqti-lwt-unix` around.. But had to guess why I could not get things working and why my LSP complained of unbound modules).

07:31 <discocaml> <.pseud.>

07:31 <discocaml> <.pseud.> Then I finally looked at their example since I could not, for the life of me, find examples anywhere else and since the https://ocaml.org/p/caqti-lwt/1.9.0/doc/Caqti_lwt/index.html was what I found and when I selected 2.2.4 I saw nothing (turns out 2.1.1 has an entry). And so on and so on and so on.

07:31 <discocaml> <.pseud.>

07:31 <discocaml> <.pseud.> If one is well-acquainted with the ecosystem and language, maybe reading the source is enough, maybe you'll know if there's API docs elsewhere and how to build them. Maybe you'll quickly guess that you should require `caqti-lwt.unix` and so on.. But it's just unfortunate that the library one is lead toward is also incidentally a much more difficult-to-use and under-documented thing.

07:31 <discocaml> <.pseud.>

07:31 <discocaml> <.pseud.> ((I get why, in the abstract.. I am assuming the author is trying to leverage the type-system to provide checks and guarantees beyond what e.g. the base sqlite3 library does. But the point still stands.))

07:34 <discocaml> <regularspatula> Yeah I understand where you’re coming from. I’m not sure how to turn your critique into actionable things though. Potentially posting on the discus forum?

07:34 <discocaml> <regularspatula> At least more people could see the issues there

07:36 <discocaml> <.pseud.> I guess for me it's more of a getting it off my chest thing 😄

07:36 <discocaml> <.pseud.> Yea.. I do understand, it's not that helpful if not turned actionable.

07:36 <discocaml> <.pseud.>

07:36 <discocaml> <.pseud.> I kind of expect a parade of people dismissing it and calling it "skill issue" though. And I get it, I'm not the PM of whoever writes caqti. I, and anyone else stumbling across that library, is owed nothing.

07:36 <discocaml> <.pseud.> So it feels useless to raise, really.

07:37 <discocaml> <regularspatula> I don’t think it’s useless. There have been posts like this before. But of course it takes some activation energy to write up a post which isn’t trivial to overcome either

07:38 <discocaml> <regularspatula> You mention building api docs on your own, there is odig for that. It will give you local docs for the versions in your switch

07:41 <discocaml> <.pseud.> Yea. I suspected there would be. But when you're just spinning your wheels and trying to write something small to see if you can actually make ocaml do something for you (so.. investigating the worth of investigating further), it's already a lot.

07:41 <discocaml> <.pseud.>

07:41 <discocaml> <.pseud.> Like I said, this is not unique to ocaml. It can be a pain to figure out how to do "modern" .NET/F# projects, or getting started with Clojure (which project management tool should I use) etc etc etc.

07:41 <discocaml> <.pseud.> So by the time I had finally wrapped ocaml up in a nix development shell (to provide system libraries) and, sort of (ish) figured out how to make dune do what I wanted.. a lot of time was spent and I was more interested in writing some code 😄

07:41 <discocaml> <.pseud.> .

07:41 <discocaml> <.pseud.> But it's all good. I'll use the "dumb person" sqlite library. It maps closely to what I'm used to and I won't know why I'm missing out anyway, so it's all good 😄

07:42 <discocaml> <regularspatula> Yeah it’s unfortunate it’s not a ln easier path to coding

07:44 <discocaml> <regularspatula> You mean this one https://github.com/mmottl/sqlite3-ocaml

07:44 <discocaml> <.pseud.> yup

07:45 <discocaml> <regularspatula> Caqti uses that one if I recall

07:45 <discocaml> <regularspatula> It’s good

07:46 <discocaml> <.pseud.> it seems simple enough that I can read the type-defs and API docs and have SOME notion of how to put the lego pieces together to actually do stuff. So that's a plus 🙂

13:08 <discocaml> <contificate> Honestly, there's so little that any idea would effectively be the first tutorial-esque video about that topic.

13:10 <discocaml> <contificate> You can connect a bunch of stuff together as well, anything using Lwt may be a networked program requiring incremental (buffered) parsing of some format (say, using Angstrom.Buffered), maybe even using JSON at some point as well (Yojson, maybe even derived with ppxes).

13:44 <discocaml> <dubious245> The only decent Ocaml content is the Cornell 3110 videos. Which are frankly great.

13:45 <discocaml> <dubious245> But thats covers a limited range of ocaml.

13:45 <discocaml> <dubious245> Everything else is 50 minutes conference talks where they waffle for an hour.

15:52 <discocaml> <gabyfle> Hello, is there a particular reason that could lead `caml_ba_alloc` to block ?

16:08 <discocaml> <gabyfle> I have a code where using FFI, I read/write to/from bigarrays. "read then write" don't cause any blocking but "write -> read" cause caml_ba_alloc to block and I can't find one. In my write function, I don't do anything that alter OCaml heap since I'm just reading values from the memory and writing them into an audio file

16:09 <discocaml> <gabyfle> Here is my code for the writing <https://pastebin.com/cEKBKANZ>

16:10 <discocaml> <gabyfle> Here is my code for the writing <https://pastebin.com/cEKBKANZ> (it's quite very simple code actually, it doesn't do anything fancy)

17:08 <discocaml> <youngkhalid> I really enjoyed this one, I wish Tsoding did more OCaml videos :

17:08 <discocaml> <youngkhalid> https://www.youtube.com/watch?v=iPrltr1b6Q4

21:43 <adrien> without really reading the code, maybe the alloc needs to trigger a collection first and you may deadlock