00:43 <rgrinberg> companion_cube: have you checked out csexp?

00:44 <companion_cube> not really, but it's not human writable, is it?

00:45 <rgrinberg> Yes it’s not meant for that. It reminds of our work on bencode however

00:54 <companion_cube> yes indeed!

00:55 <companion_cube> these days I like resp3 for some things

00:56 <companion_cube> or msgpack if it's binary

01:28 <rgrinberg> one nice thing about csexp is the simplicity. it's a single module you can drop in anywhere

01:28 <companion_cube> grrrrrrr, ocaml-protoc doesn't like `required`?

08:42 <Leonidas> companion_cube: ocaml-protoc had so many issues we wrote *two* replacements

11:24 <flux1> I haven't used ocamlc-protoc, but it has probably the same problem as other protobuf-converters: it uses protobuf

11:29 <flux1> is it a feature of Sedlex that the syntax `match%sedlex` seems to work only in the top-level of functions?

11:32 <flux1> actually the rule seems to be "inside a let"

11:33 <flux1> or at least "not inside a pattern matching without being inside a let" :)

12:28 <theblatte> in a dune file for a sub-library of my project, I would like to add a (preprocessor_deps (file ../foo)) but dune cannot find "foo" even though it's the correct path from the project root

12:28 <theblatte> if I put foo in the same directory as my library then dune is happy but I'm not

12:29 <theblatte> I'm guessing I need to give up and maybe use (copy_files ) so dune can pretend files are where it wants it but is there another way?

12:32 <theblatte> oh, it's my fault: I didn't add the directory where "foo" is to the (dirs ) stanza at the root of my project

14:11 <match_it> hi all. I'm beginner with ocam. I receive an error compiling a package. The error concerns GCC and I would switch to G++. Where can I set default system compiler?

14:11 <match_it> thanks

14:16 <malc_> match_it: ocamlc -help | grep -- -cc | head -1

14:16 <malc_> -cc <command> Use <command> as the C compiler and linker

14:17 <match_it> malc_, thanks, let's see

18:19 <companion_cube> Leonidas: which replacements?!

18:21 <companion_cube> (I see no other program that takes a .proto and generates code from it, at least)

18:50 <asoni> companion_cube: I think it is https://github.com/issuu/ocaml-protoc-plugin

18:52 <companion_cube> ah but this uses the official protoc, right?

19:41 <flux1> for some reason `Lwt_main.run` + `Cohttp_lwt_unix.Client.post` spins my CPU at 100% while an `Lwt.pick` + `Lwt_unix.timeout` -mechanism cannot stop it from running. any ideas?

19:48 <flux1> hmm, OUnit2 has support/perhaps uses lwt internally, so perhaps that's they key?

19:48 <flux1> ..but the funny thing is that I saw it working a few times

20:10 <flux1> using the lwt_wrapper from ounit2-lwt seems to have resolved it

20:10 <flux1> quite a stupid error mode though :/

20:59 <higherorder> Hello! I am new to Ocaml and I have a basic question on modules. Is it common to have one module per type? For example if I am modelling an AST, would it make sense to have one module for each type of node in the AST, with an abstract type for the node and a function to construct a node of that type?

21:00 <def> no

21:00 <def> mutually recursive types are almost always defined in the same module.

21:04 <higherorder> def: ok; thanks!

22:30 <Leonidas> companion_cube: https://github.com/issuu/ocaml-protoc-plugin and https://github.com/martinslota/protocell

22:30 <Leonidas> companion_cube: yes, this uses the official protoc which has the advantage of actually implementing the parser correctly.

22:47 <companion_cube> what are the issues with ocaml-protoc?