11:58 <discocaml> <xavierm02_> Is it possible to organize several independent files containing expect tests in such a way that they all share the same dune file and it is possible to run the expect tests of a single file?

11:58 <discocaml> <xavierm02_> If I put all the files in the same directory, then I don't know how to tell dune to only run the expect tests of one of the files; if I put them each in their subdirectory then I can do `dune runtest lib/subdir` but I need to have a dune file in each subdir because otherwise ppx_expect_runtime raises an exception telling me it didn't find the file in `lib`...

13:52 <discocaml> <bluddy5> Arise! https://discuss.ocaml.org/t/monadic-library-for-eio-capabilities/14202/38

13:54 <companion_cube> 😂

13:57 <olle> Arise what?

13:59 <discocaml> <bluddy5> (undead thread)

13:59 <companion_cube> lfg

14:00 <discocaml> <bluddy5> How many people are actually actively using eio? Aren't most people still on lwt?

14:00 <companion_cube> I'm not entirely sure

14:00 <companion_cube> I'm on OCaml 5 but without Eio

14:01 <olle> Ah

14:02 <discocaml> <bluddy5> lwt has 613 uses on opam. eio has 28.

14:19 <discocaml> <JM> The last eio adoption I've seen (smaws), if I read it right, sticks the entire `env` in a context so capabilities don't appear in the API at all.

14:19 <discocaml> <JM> That seems the only sensible way to have a good dev experience when building on top of eio 🤷‍♂️

15:23 <discocaml> <lukstafi> Any idea where in Landmarks a discrepancy between `Time` (https://en.wikipedia.org/wiki/Time_Stamp_Counter) and `Sys time` could be coming from? It's for `ROOT`, could it be that `Landmark.reset` does not reset the sys timer? I have a loop that should be doing the same thing, `Time` has a very slight statistical upward drift, while `Sys time` grows monotonically, something like `delta0 + 0.45 * t * delta0`.

15:25 <discocaml> <lukstafi> (I'll check if it comes from condition variables.)

15:25 <companion_cube> sys time is cpu usage, isn't it?

15:27 <discocaml> <lukstafi> `sys_time : bool;

15:27 <discocaml> <lukstafi> Also collect Sys.time timestamps during profiling.`

15:27 <discocaml> <dinosaure> `Sys.time` is (depending on your system however) probably the `times` function

15:27 <discocaml> <dinosaure> (see `man 2 times`)

15:28 <discocaml> <dinosaure> if it's really sensible, it's probably better (for micro-benchmark) to use a monotonic clock like `clock_gettime`

15:29 <discocaml> <dinosaure> (and if you want to be really precise, `rdtsc` is probably the best)

15:31 <discocaml> <dinosaure> (and I reiterate, if it's a question of calculating execution time, `landmarks` may not be the best, it's very good for a GC view but `bechamel` is probably preferable for micro-benchmark).

15:34 <companion_cube> use mtime for this

15:34 <companion_cube> monotonic time

15:37 <discocaml> <dinosaure> yes, but the metric is still volatile. you need to do an extra analyze (like ols or ransac) to have a good confidence about results

15:43 <discocaml> <lukstafi> Alright! mtime resolves it, thanks! `Sys time` isn't getting reset.

15:44 <companion_cube> on linux at least, mtime is super fast, it bypasses the system call entirely

15:44 <companion_cube> vdso or whatnot

16:53 <Mister_Magister> Hello beautiful people, i've decided to finally start learning ocaml and couple paragrpahs into tutorial i am already confused. Tutorial states that let makes a constant but then couple lines above it was adding value to array that was defined with let, which would mean it is not constant but mutable, and at this point i am confused

16:58 <f[x]> name bound by let is constant - as in points to the same value, value itself can be mutable

16:59 <discocaml> <yawaramin> sorry can you point us to what paragraph exactly you are reading, it will help answer your question better. the exact link would be good

17:00 <discocaml> <._null._> The definition of constant and variable is a bit blurry since they are seen as synonymous in OCaml : you can't change the value a variable contains, but if a value is mutable then it can be mutated

17:00 <octachron> An alternative viewpoint is that "let" gives a name to an expression. The name is constant and will always refer to the same object. But the contents of the named object may change.

17:06 <Mister_Magister> @yawaramin https://ocaml.org/docs/tour-of-ocaml#expressions-and-definitions

17:07 <discocaml> <yawaramin> sorry, where is it adding value to array that was defined with let? i don't see any mention of arrays here

17:08 <Mister_Magister> list if we want to be pedantic

17:09 <discocaml> <._null._> Lists aren't mutable, no list is being edited here

17:10 <Mister_Magister> it's literally doing # 9 :: u;;

17:10 <discocaml> <._null._> This is not changing u

17:10 <Mister_Magister> oh… it doesn't edit the list

17:10 <Mister_Magister> ohhh

17:10 <discocaml> <._null._> This is a construiction

17:10 <Mister_Magister> now that makes sense

17:10 <Mister_Magister> sanity restored

17:10 <discocaml> <yawaramin> it's kinda important to differentiate between lists and arrays in OCaml

20:53 <discocaml> <chrisarmstrong> yes, and I wouldn't argue it's the most correct usage (I'm the author 😂 ). In that case, it's a long-lived context i.e. a connection pool, so I need to keep the same env around