03:10 <discocaml> <idontwantausernamefeckoff> if i have a value of type `[ ``A | ``B ] t`, how would i pass this to a function expecting `[ ``A ] t`

03:11 <discocaml> <Kali> as a discord formatting tip, you can use \`\`code\`\` to prevent needing to use double ` for polymorphic variants

03:11 <discocaml> <idontwantausernamefeckoff> oh thank you

03:11 <discocaml> <Kali> example: ``[ `Like | `This ] ``

03:12 <discocaml> <idontwantausernamefeckoff> i'm really confused by polymorphic variants in general

03:14 <discocaml> <Kali> this is not valid: if the value can be `` `A`` or `` `B``, how would you pass it to something that only accepts `` `A``?

03:15 <discocaml> <idontwantausernamefeckoff> hm

03:15 <discocaml> <idontwantausernamefeckoff> https://ocaml-multicore.github.io/eio/eio/Eio_unix/Stdenv/index.html#type-base

03:17 <discocaml> <idontwantausernamefeckoff> thats the particular type. caqti takes ```[`Generic] Eio.Net.ty Eio.Std.r```

03:24 <discocaml> <darrenldl> which caqti function?

03:25 <discocaml> <idontwantausernamefeckoff> manually editing it to take a ``[`Generic | `Unix`` value connects to postgres fine, and i'm already having to manually reconstruct the env object so i'm wondering if this is a case of eio changing the api (i had to pin to master to run with 0.14) or im missing something

03:26 <discocaml> <idontwantausernamefeckoff> connect_pool

03:27 <discocaml> <idontwantausernamefeckoff> manually editing it to take a ``[`Generic | `Unix`` value connects to postgres fine, and i'm already having to manually reconstruct the env object so i'm wondering if this is a case of eio changing the api (i had to pin to master to run with 0.14) and i should open an issue or im simply missing something

08:23 <dh`> blurgh I hate it when your lists get too long for List.map

08:29 <discocaml> <geoff> Not a problem anymore with the new stdlib (with tail mod cons) though, right?

08:52 <dh`> probably not but this is not 5 yet

09:30 <reynir> When using caqti, do you know of an example where you don't pass `Caqti_type.unit` as the second argument to `Caqti_request.Infix.( -->. )` ?

09:35 <reynir> So the type is `val Caqti_request.Infix.( -->. ) : 'a Caqti_type.t -> unit Caqti_type.t -> _` so unless you `Caqti_type.custom` it only ever makes sense to pass `Caqti_type.unit` (and I can't really think of a case where a custom type makes sense)

13:31 <neiluj> hey! got a typing problem: https://github.com/jtcoolen/ocaml-pari/blob/main/examples/kzg.ml#L115, here the labeled arg n is supposed to be of type Integer.t=(integer, ring) typ, the type of residue_class is (finite_field, field) typ -> (finite_field, ring) typ Polynomial.t and Polynomial.( .%[] ) : ('a, ring) typ t -> int -> ('a, ring) typ but here the type of .%[] is (finite_field, ring) typ Integer.t Polynomial.t -> int -> Integer.t,

13:31 <neiluj> while it should be (finite_field, ring) typ Polynomial.t -> int -> (finite_field, ring) typ

13:32 <neiluj> is there an explanation for that?

17:03 <marcc> Just noticed that this PR doesn't seem to be getting merged and contains a fix I requir: https://github.com/roddyyaga/pg_query-ocaml/issues/16 :(

17:03 <marcc> Is the easiest way just to fork it and apply the patch and have it locally instead?

20:01 <discocaml> <idontwantausernamefeckoff> marcc: yeah

20:01 <discocaml> <idontwantausernamefeckoff> dune vendoring is your friend

21:50 <Anarchos> how could i break the type circularity ? https://bpa.st/CWYA

21:51 <discocaml> <._null._> and ?

21:51 <discocaml> <._null._> (the keyword)

21:59 <Anarchos> ._null_. demonstration > step > theorem > demonstration ...

22:02 <discocaml> <._null._> `type demonstration = (formula list * step) list and theorem = (formula, demonstration) theorem_logic and step = (formula,theorem) step_logic`

22:03 <discocaml> <._null._> Basically, there is no need to break the cycle, it is supported

22:06 <Fardale> Anarchos: you either need to compile with -rectypes or you need to introduce a constructor, for exemple having `type step = Step of (formula, theorem) step_logic`

22:07 <discocaml> <._null._> Oh woops

22:08 <discocaml> <._null._> Add `[@@unboxed]` after the type declaration and it will compile as if you had recursive types

22:18 <Anarchos> thanks