00:25 <companion_cube> Haven't looked in a while but isn't there a pipe option?

00:59 <dmbaturin> companion_cube: I have a moonpool question. Suppose I have a list of 100 file names and I want, WLOG, two threads to read them in parallel. If I wrap the reading function into `spawn`, do I need to do anything special about the IO functions, or stock functions from Sys/Unix are fine to use? I'm sorry if this a stupid question, I'm still trying to build a mental model of the multicore runtime in general.

00:59 <dmbaturin> The parallel for from domainslib is not safe to use that way, it led to unhandled effect errors when I tried.

00:59 <companion_cube> It'll be just fine if you use the blocking IO API

01:00 <companion_cube> I mean I'd make a pool and dispatch one task per file

01:00 <companion_cube> Possibly a parallel map, even

01:02 <dmbaturin> Possibly even dumber question. Is there a specific reason why there's no ready to use `'a list -> 'b list` parallel map there in moonpool? Or there actually is and I'm missing it?

01:06 <companion_cube> Iirc there's one in the Fut module

01:07 <companion_cube> Not exactly that signature though, but with ocaml 5 you can also just do a List.map with spawn inside, followed by a List.map with await

01:08 <dmbaturin> Oh, `Fut.map_list`.

01:11 <companion_cube> But also, really, two successive List.map can do

01:11 <companion_cube> That's the beauty of await

01:12 <companion_cube> (it just means the `List.map await futures` has to run in a effect-handling context)

21:24 <discocaml> <dubious245> Just wanted to vent but why does this research papers and all its implementation all insist on naming everying like (i, j) (i2,j2) (i1, j1) (i3,j3) (i4,j4)

21:24 <discocaml> <dubious245>

21:24 <discocaml> <dubious245> I spent half the day reading this paper and its implementation to figure out what its doing with all this i1, j1 = i2, j3 and j2, j2 = i4, j4.