Anton Bachin
Anton Bachin
That does seem wrong @Lupus. I checked the standard `Unix`, and it looks like the Lwt code is directly based on it: https://github.com/ocaml/ocaml/blob/d14b3ec19f44fc4ac88c1fa4b7c97e201c689b96/otherlibs/unix/getaddrinfo.c#L117. It also doesn't have an `else` branch....
Well, one of the most immediate issues is that the `EAI_*` error codes are not included in [`type Unix.error`](https://caml.inria.fr/pub/docs/manual-ocaml/libref/Unix.html#TYPEerror).
They do, however, appear in the POSIX man pages for the functions, e.g. https://man7.org/linux/man-pages/man3/freeaddrinfo.3p.html#ERRORS. So this could be an argument for adding them (eventually?). They are also present in libuv...
In the near term, we could map all of them to `Unix.EUNKNOWNERR` from `Unix.error`.
Regarding Discord, another option is to remove it instead. I put it in there because I monitored Discord.
Thanks for the report! I've adjusted the docs (see the attached commit). We actually need to rewrite the `Lwt_unix` docs completely, they are very much inadequate. For a Windows implementation...
See also #269, in which @yallop added support for choosing the backend.
libev should probably also be chosen by default on mac: https://github.com/ocsigen/lwt/blob/ee466e23979644beb30f2d51a9835f5ffe34a3ab/src/unix/config/discover.ml#L860-L868
This also seems to be a duplicate of #87.
@Lupus that's right. I couldn't remember all the usual benefits :) the libuv API makes it easy to run multiple event loops, one per system thread. It's not quite thread-safety...