17:30 <leon-p> just went hunting for what I thought was a zig compilation because somehow a cursor_shape value was negative after getting it via inline else and @field(@tagName()). Turned out that struct was never initialized, which I missed because other members of it had default values set... fun :)

19:23 <ifreund> leon-p: valgrind probably woulda caught it :P

19:24 <leon-p> ifreund: it actually didn't, linking with chicken messes up the debug info

19:26 <ifreund> :/

19:29 <leon-p> yeah, it makes debugging fun. Last week I had a hard time even noticing a bug, because chicken was catching the SIGSEGV raised by an assert and turning that into a runtime scheme error

19:30 <ifreund> weird

19:30 <ifreund> Debugging my janet stuff has been pretty easy so far

19:31 <ifreund> gdb works pretty much as expected

19:31 <ifreund> valgrind also worked great to get all the leaks sorted

19:32 <ifreund> apparently there's also debugging stuff implemented in Janet for debugging janet code but I haven't needed to learn how to use those yet

19:33 <ifreund> I must say, this single threaded cooperative multitasking stuff is very comfy

19:34 <ifreund> for example, I can just do this and spawn a process without a zombie and without blocking the event loop: https://codeberg.org/ifreund/rijan/src/commit/25938e0ded6d42763d455a0164c04b82098825e0/src/seat.janet#L21-L22

19:35 <leon-p> is that basically janets version of async?

19:36 <ifreund> yeah I think this is what people usually mean by async/await

19:37 <ifreund> cooperative multitasking is more precise though :D

19:38 <leon-p> in what way? I have never really done anything with async

19:41 <ifreund> leon-p: as I understand it, what janet does isn't really the same as javascript async await. There's no special syntax and no function coloring

19:41 <ifreund> (RE: function coloring https://journal.stuffwithstuff.com/2015/02/01/what-color-is-your-function/)

19:44 <ifreund> janet the language has "asymmetric stackful coroutines" if I've got my terminology right

19:44 <ifreund> and the janet standard library uses this language feature to make a nice comfy event loop

19:48 <leon-p> my understanding of async has always been that it's usually just syntax sugar around event loops, especially w.r.t. events raised by your own code

19:48 <leon-p> scheme uses continuations for single-threaded multitasking, but I have not messed with the chicken version of it, only tried the guile fibres library

19:50 <ifreund> leon-p: that's pretty much what this is, janet even calls them fibers

19:50 <ifreund> continuations can be used to implement coroutines pretty trivially

20:27 <pkap> any idea why this does not compile https://codeberg.org/pkap/river/src/commit/a6a8a4612ad6da4b57af2871f9865e91f49e8f62/river/Server.zig#L564 with "unable to resolve comptime value"

20:28 <pkap> there's no comptime in these types as I understand it.

20:28 <ifreund> pkap: that * shouldn't be there

20:29 <ifreund> it's parsing the left hand side of that = as a type

20:29 <ifreund> and types must be comptime known

20:34 <pkap> ah yes of course. Well, it's late. Thanks ifreund

20:34 <ifreund> no problem!

20:34 <ifreund> the error message could probably be better

20:35 <pkap> Maybe soon done with the toplevel capture stuff. Even though it has to wait until the next wlroots release.

20:36 <ifreund> well, you can probably use it yourself even if it has to wait to get merged :D

20:38 <pkap> Yes I totally will. But I also need this in order to use it for my use cases: https://github.com/emersion/xdg-desktop-portal-wlr/pull/329

20:45 <pkap> is there a reason the in Server.zig, the loop is not a field? https://codeberg.org/pkap/river/src/commit/a6a8a4612ad6da4b57af2871f9865e91f49e8f62/river/Server.zig#L136

20:48 <ifreund> pkap: I don't like redundant state, libwayland already stores a pointer to the event loop, why should river store a second one?

21:00 <pkap> Gotcha

21:03 <pkap> also this wouldn't help me. My problem is how to access the wl_server field in a static function (fn handleNewForeignToplevelCaptureRequest)

21:04 <pkap> it takes *wl.EventLoop as an argument.