09:21 <Anarchos> According to https://pubs.opengroup.org/onlinepubs/9699919799/, sendto should fail with ENOENT if pathname is "". This is precisely the case in the recvfrom_unix unit test. So i think the 'OK' reference is incorrect.

09:27 <Anarchos> Should i open an issue on github.com/ocaml ?

09:39 <octachron> There is no "" pathname in sight in this test? `sendto` is called on an anonymous socket.

09:42 <octachron> (the testsuite is run on linux, freeBSDs, openBSDS, macos and omniOS, it is unlikely to contain linuxism)

09:56 <Anarchos> the socket is created on ""

09:57 <Anarchos> octachron by anonymous socket you mean the socket is not bind to an address.

10:03 <Anarchos> octachron but i understand this line is still using an empty pathname :

10:04 <Anarchos> test_sender ~client_socket ~server_socket ~server_addr ~client_addr:(ADDR_UNIX "")

10:04 <octachron> The socker is created with "Unix.socket PF_UNIX SOCK_DGRAM 0" and is not bound to any address.

10:05 <octachron> The client address is only used for comparing it to the path returned by recvfrom

10:06 <Anarchos> octachron ok i will look into my netswork stack

10:38 <jiquiame> Hey, I'm having trouble setting up dependancies for a project using js_of_ocaml. Basically, opam on windows has only access to 4.0.0, which is incompatible with --disable option, which I need to provide in linux/osx platforms with newer versions of the package. Using dune variables, I'm able to specify the jsoo compilation flags according to os_type, however, that does not make sense with the

10:38 <jiquiame> dependancy constraints I set in the dune-project. I would rather send flags according to the jsoo version, but even if dune provides %{version:js_of_ocaml}, I would then need to list each version of jsoo to specify the flags as I couldn't find a way to make a comparison based on this information. I would gladly take tips on that

10:40 <jiquiame> I also have trouble finding out how to specify specific versions of libraries in executables packaged with opam, especially that I'd also need to do it with allowing 4.0.0. Maybe that would be different opam packages, defined differently in dune?

22:08 <discocaml> <functionalprogramming> is there a popularly-used "double-ended" list for OCaml?

22:10 <discocaml> <functionalprogramming> it seems i can use two lists, will try that

22:10 <discocaml> <Kali> as in, a doubly linked list?

22:11 <discocaml> <functionalprogramming> sure, i mean i guess my q is, is there a library for https://discuss.ocaml.org/t/how-to-represent-a-double-ended-linked-list/13354/2?u=mizlan

22:11 <discocaml> <functionalprogramming> i dont want to rewrite all the functions myself

22:13 <discocaml> <functionalprogramming> i suppose i can write them

22:15 <masterbuilder> @functionalprogramming what do you need it for? what API do you wish for etc.

22:16 <discocaml> <Kali> oh, not a doubly-linked list, an actual two-ended list

22:19 <discocaml> <functionalprogramming> i wish to use it for constant-time "rotating" one element to the end, as well as O(M)-time appending (to end) of a length-M list

22:20 <discocaml> <functionalprogramming> although i suppose my N/M is never very big, so I might just use one list and use @ [...]

23:52 <discocaml> <hockletock> `opam search deque` shows that fmlib has a deque implementation

23:52 <discocaml> <hockletock> api docs here https://hbr.github.io/fmlib/odoc/fmlib_std/Fmlib_std/Deque/index.html

23:55 <discocaml> <contextfreebeer> containers has `CCFQueue` too