<< 18-12-2023 >>
00:12:45FromDiscord<Elegantbeef> Then what is the point of adding?
00:18:48FromDiscord<System64 ~ Flandre Scarlet> In reply to @Elegantbeef "Then what is the": FM synthesis?↵It consists to add a value previously generated by an oscillator to next oscillator's phase accumulator
00:22:16FromDiscord<Elegantbeef> So then you do not actually want to add
00:22:41FromDiscord<Elegantbeef> So sure cast is the only way to achieve what you want
00:27:25FromDiscord<griffith1deadly> beef, if i have big ref object (big size from object fields) i'm also need declare field object types like ref?
00:27:48FromDiscord<System64 ~ Flandre Scarlet> In reply to @Elegantbeef "So then you do": I want to add a value to the phase accumulator
00:28:55FromDiscord<Elegantbeef> Idk
00:52:54FromDiscord<yoyojambo> sent a code paste, see https://play.nim-lang.org/#ix=html>
00:54:20FromDiscord<yoyojambo> Is it the tuple? I couldn't find what it could refer to that the compiler dislikes about it.
00:54:36FromDiscord<Elegantbeef> `Positive` is a sub range type so should be explicitly init'd
00:55:16FromDiscord<yoyojambo> Ahhhh of course, making it an `int` solves it, thanks!
00:57:25FromDiscord<us> can you talk some details?
01:05:07*Mister_Magister quit (Quit: bye)
01:07:02*Mister_Magister joined #nim
01:22:13FromDiscord<Elegantbeef> @griffith1deadly to elaborate what you should do depends on what you're doing, speaking in generalities does not generally benefit 😄
01:22:30FromDiscord<Elegantbeef> As such I'll say those objects should be only ref if you are moving them around a lot
01:23:19*krux02 quit (Quit: Leaving)
01:23:44FromDiscord<Elegantbeef> Otherwise you do not benefit much from having them as `ref`s and just introduce a "mistake"
02:12:45FromDiscord<demotomohiro> In reply to @sys64 "``(osc.pAcc + prevSample)`` I": If you want to do such addition without overflow, convert both numbers to int32 and add.↵Then, convert result to uint16 or int16.
02:47:35FromDiscord<griffith1deadly> In reply to @Elegantbeef "<@620135501201932300> to elaborate what": it's like array[256, array[64, array[64, int32]]], but just like container's with smaller size and done with seq
02:48:13FromDiscord<griffith1deadly> and it's object should be in memory in big counting, like 1000-5000
02:48:35FromDiscord<Elegantbeef> Right but many instantiations do not really mean "use ref"
02:49:05FromDiscord<griffith1deadly> it also need to be modified from many sources
02:49:05FromDiscord<Elegantbeef> If you can move resources instead of copying them, or would benefit from a pointer
02:49:40FromDiscord<Elegantbeef> If it needs to be viewed from multiple places you need a reference then
02:51:26FromDiscord<griffith1deadly> In reply to @Elegantbeef "If you can move": like object poll or?
02:52:01FromDiscord<Elegantbeef> You can either use a pool with a 'pointer' which is an index into it, or use references
02:52:58FromDiscord<griffith1deadly> (edit) "poll" => "pool"
02:53:01FromDiscord<griffith1deadly> thanks!
02:53:10FromDiscord<Elegantbeef> The first is a bit more tedious as you have to pass the environment around
02:53:38FromDiscord<Elegantbeef> It also is faster with smaller objects than you have, but is probably on par with `ref T`
04:02:34*xutaxkamay quit (Ping timeout: 255 seconds)
04:16:13*xutaxkamay joined #nim
04:49:23*Qaziquza quit (Ping timeout: 252 seconds)
05:54:29*jmdaemon joined #nim
07:34:43*advesperacit joined #nim
08:02:30FromDiscord<ieltan> Writing a bunch of higher order functions for `Option` à-la Rust and wondering if some parameters should be `sink`
08:03:49FromDiscord<ieltan> I guess they should be hm
08:05:16FromDiscord<Elegantbeef> Any parameter that can be moved, should be annotated sink
08:07:41FromDiscord<ieltan> Alright 🙂
08:36:55*PMunch joined #nim
09:01:43NimEventerNew thread by mantielero: Scinim - how to contribute a package, see https://forum.nim-lang.org/t/10780
09:15:26FromDiscord<Phil> @vindaar @hugogranstrom It might make sense to enable "Discussions" for the SciNim on github organization.↵mantielero is a useful reminder here that Discussion Pages for an organization can be a decent way for members of the community to get in contact with you guys.↵Not everyone is in the forum or on this discord 😉
09:15:50FromDiscord<Phil> (edit) "you guys.↵Not" => "it.↵Not"
09:25:11FromDiscord<vindaar> Ah, good idea! Will take care of it
09:25:59*derpydoo quit (Ping timeout: 260 seconds)
09:33:21FromDiscord<ca_lobo_14153> sent a long message, see <!doctype html>
09:40:17FromDiscord<demotomohiro> In reply to @ca_lobo_14153 "My first time playing": `stripLineEnd` doesn't return string. It modify the argument.
09:42:37FromDiscord<demotomohiro> !eval import strutils; var r = "FooBar\n"; r.stripLineEnd; echo r & ".txt"
09:42:42NimBotFooBar.txt
09:59:08FromDiscord<ca_lobo_14153> aaaaah, ok so the documentation is incorrect as using a var to capture the osproc.execCmdEx results and then applying stripLineEnd turns it into a void vs actually modifying the results. I've figured out that assigning the result[0] to another variable, allows me to avoid using stripLIneEnd entirely...
09:59:38FromDiscord<System64 ~ Flandre Scarlet> In reply to @demotomohiro "If you want to": Does it clamp or overflow when converting back to uint16?
10:03:27FromDiscord<odexine> overflow
10:03:56FromDiscord<demotomohiro> In reply to @sys64 "Does it clamp or": !eval var x = 65536.int32 + 7.int32; echo x.uint16
10:07:02FromDiscord<sOkam! 🫐> sent a code paste, see https://play.nim-lang.org/#ix=html>
10:07:35FromDiscord<System64 ~ Flandre Scarlet> In reply to @demotomohiro "!eval var x =": Looks like the bot is dead
10:08:30FromDiscord<System64 ~ Flandre Scarlet> !eval echo "test"
10:08:33NimBottest
10:09:02FromDiscord<System64 ~ Flandre Scarlet> sent a code paste, see https://play.nim-lang.org/#ix=html>
10:09:50FromDiscord<demotomohiro> It still alive.↵Anyway, `(65536.int32 + 7.int32).uint16 == 7`
10:11:07FromDiscord<System64 ~ Flandre Scarlet> !eval echo(255.uint16 + (-1).uint16)
10:11:10NimBotCompile failed: /usercode/in.nim(1, 23) Error: -1 can't be converted to uint16
10:11:54FromDiscord<System64 ~ Flandre Scarlet> !eval echo(255.uint16 + cast[uint16](-1))
10:11:57NimBot254
10:12:12FromDiscord<System64 ~ Flandre Scarlet> Oooh!
10:13:03FromDiscord<System64 ~ Flandre Scarlet> !eval echo(255.uint16 + cast[uint16](-255))
10:13:06NimBot0
10:13:29FromDiscord<System64 ~ Flandre Scarlet> And finally
10:13:41FromDiscord<System64 ~ Flandre Scarlet> !eval echo(255.uint16 + cast[uint16](-256))
10:13:44NimBot65535
10:13:58FromDiscord<System64 ~ Flandre Scarlet> Makes sense
10:16:13*Qaziquza joined #nim
10:17:01FromDiscord<ca_lobo_14153> In reply to @heysokam "<@1164001202451775498> tip: writing": Thanks sOkam! I'd forgotten about that...
10:18:53FromDiscord<ca_lobo_14153> (edit) "In reply to @heysokam "<@1164001202451775498> tip: writing": Thanks sOkam! I'd forgotten about that... ... " added " fixed"
10:39:22*Qaziquza quit (Ping timeout: 260 seconds)
11:09:14*jmdaemon quit (Ping timeout: 260 seconds)
11:34:51*koltrast quit (Ping timeout: 256 seconds)
11:46:36NimEventerNew thread by coneonthefloor: How to await a list of async procedures, see https://forum.nim-lang.org/t/10781
12:05:10*koltrast joined #nim
12:10:28FromDiscord<rakgew> sent a code paste, see https://play.nim-lang.org/#ix=html>
12:17:48FromDiscord<rubythulhu> im getting real “cool there’s a nimble package for X” -> “breaks during nim install, written for a way old version of nim, current version’s dependencies break things somehow” kinda vibes lately
12:18:59FromDiscord<Phil> hm?
12:19:30FromDiscord<Phil> If this is a package that does DB interaction then yeah, 2.0 kinda killed you with how it moved the packages around.↵Then again, that was a breaking version change so to be expected
12:48:59FromDiscord<rubythulhu> its a whole bunch of stuff
12:52:04FromDiscord<rubythulhu> i've had that kind of thing happen a bunch, whichever aws sdk thing felt most actually workable, and how i cant even get the nim godot stuff to work, etc
12:52:48FromDiscord<rubythulhu> kind of to be expected for a younger / less popular language but, is a bit disheartening
13:08:03FromDiscord<nnsee> In reply to @rubythulhu "i've had that kind": this ofc doesn't apply to huge libraries such as nim-godot or whatever, but if a lib i'm thinking about using is small enough, or a wrapper around a web API or something like that, I usually prefer to write it myself to begin with. Partly because of the problems you mentioned, but also because I enjoy writing Nim and don't mind having an excuse to do it
13:08:44FromDiscord<rubythulhu> oh, i agree. its just trying to do something more fun than play projects
13:08:56FromDiscord<rubythulhu> nim's still the best i've found for raspberry pi pico
13:09:11FromDiscord<rubythulhu> tho i havent tried my old code under nim2 O_O
13:10:09FromDiscord<nnsee> most of my old code just worked
13:12:53FromDiscord<rubythulhu> yeah i can maintain my own code
13:13:10FromDiscord<rubythulhu> is the nimble ecosystem tho
13:13:18FromDiscord<mratsim> In reply to @rubythulhu "tho i havent tried": There is a tradeoff between trying too early while in devel, and taking focus time away to solve Nim issues, and trying too late and have a regression shipped in Nim 2.0 and then people complaining that your cold doesn't work :/
13:14:11FromDiscord<mratsim> Thankfully I have an excuse to write all the code myself with zero dependencies including the stdlib :P, so I only get compiler bugs.
13:14:15FromDiscord<rubythulhu> In reply to @mratsim "There is a tradeoff": yeah, i understand it. just getting sad how often i think "i wanna do something" -> "lets try it in nim" -> "nimble pkg" -> "oof"
13:14:46FromDiscord<rubythulhu> need to constrain my problem space more for now when choosing to play in nim
13:15:33FromDiscord<mratsim> i just assume I have to write everything myself, and I'm pleasantly surprised if I don't 😉
13:16:48FromDiscord<rubythulhu> lots of things i love it for, but i keep reaching into dark places haha
13:17:10FromDiscord<rubythulhu> the pi pico lib is my favorite way to do pico coding hands down
13:17:24FromDiscord<nnsee> what do you do with your pico
13:17:27FromDiscord<rubythulhu> and write little whatsits here and there for fun
13:17:27FromDiscord<nnsee> out of curiosity
13:17:51FromDiscord<rubythulhu> well, i haven't gone back to the project in a bit haha 😅
13:17:59FromDiscord<rubythulhu> i have a bunch of them
13:18:15FromDiscord<rubythulhu> and i watched and did ben eaters whole 6502 series
13:18:31FromDiscord<rubythulhu> and i wanted to play w/ more
13:19:30FromDiscord<rubythulhu> so my plan is to make a 65c816-based single-game arcade console, but basically use pico's to fake old hardware for bunches of it
13:22:04FromDiscord<nnsee> neat
13:22:45FromDiscord<rubythulhu> so the thing i was in the middle of before work distracted me was, i was building basically a storage/memory/bus (so basically core of the "motherboard") to communicate w/ the 65c816, and remap ram/flash and send commands to other picos over spi to control output/input/etc
13:23:48FromDiscord<rubythulhu> nim's really nice for it
13:25:39FromDiscord<rubythulhu> its hardware level and no-gc land, and it's super nice to guarantee zero cost abstractions w/ templates
13:26:22FromDiscord<rubythulhu> i tried micropython at first, but was looking at my even base running times vs what i wanted to pull off and wanted to go lower level and didn't wanna go c
13:26:29FromDiscord<rubythulhu> so i found my way into nim's lap haha
13:28:17FromDiscord<rubythulhu> plus micropython was too easy, there was already libs for everything and i wanted to write everything myself
13:34:22FromDiscord<rubythulhu> one thing i’d like to do is prototype the story of that arcade game in a more modern language and less constrained environment, i think nimwave or illwave would be cool for a tui version of it, but still too constrained to start with haha
13:36:05FromDiscord<rubythulhu> nim’s still my favorite language now, just getting frustrated when i try to use it and it turns into way more of a project than i want it to be
13:39:15FromDiscord<rubythulhu> there’s also cool stuff in the web api space i think would be fun to play with, but it’s too work-y to do while i’m looking for a job lol
13:39:19FromDiscord<nnsee> i think the C interop is sometimes an effective band-aid for that
13:39:23FromDiscord<nnsee> especially with futhark
13:39:42FromDiscord<nnsee> In reply to @rubythulhu "there’s also cool stuff": what sort of job are you looking for exactly?
13:39:46FromDiscord<rubythulhu> i haven’t played with furhark yet
13:51:22FromDiscord<rubythulhu> In reply to @nnsee "what sort of job": duno. been in devops ish stuff for about a decade, but far from the only thing i’ve done and i wonder if im getting bored with it haha
14:08:10*derpydoo joined #nim
14:24:44*sagax joined #nim
14:28:12*edr joined #nim
14:34:06NimEventerNew thread by rel: Direct C/C++ interop without generating wrappers?, see https://forum.nim-lang.org/t/10782
15:24:24PMunchrubythulhu, can highly recommend trying Futhark :)
15:25:51PMunchI use it quite often, now I'm trying out libfuse to implement a user-space filesystem for a little project and wrapping the libfuse library (with all the passC/passL stuff required) was 7 lines of code :)
15:25:55*PMunch quit (Quit: Leaving)
15:39:14FromDiscord<nnsee> "use it quite often"
15:39:20FromDiscord<nnsee> my brother in christ, you wrote it
15:40:30FromDiscord<ieltan> I'd like to contribute to futhark to add a `nep2` flag but I don't even know where to start
15:40:45FromDiscord<jviega> Yeah but I've written a lot of things I don't use often 🙂
15:41:01FromDiscord<nnsee> touché
15:41:07FromDiscord<jviega> So I appreciated the way he said it
15:41:42FromDiscord<Chronos [She/Her]> In reply to @ieltan "I'd like to contribute": Nep2?
15:44:01FromDiscord<ieltan> Uh more like `nep1` sorry
15:44:10FromDiscord<ieltan> Dunno where I got that two, I'm kinda tired lol
15:59:43FromDiscord<Chronos [She/Her]> Nep1?
15:59:48FromDiscord<Chronos [She/Her]> I don't know either aha
15:59:56FromDiscord<Chronos [She/Her]> Unless you mean the Nim naming convention?
16:01:02FromDiscord<Chronos [She/Her]> @ieltan https://github.com/Yu-Vitaqua-fer-Chronos/ngpu-bindings/tree/master/tools if you wanna look at how I made some scripts in my failed attempt to wrap WGPU for Nim, look here
16:01:09FromDiscord<Chronos [She/Her]> You can most definitely port that to Nim
16:01:43FromDiscord<bostonboston> I think he means rather than the end result looking C like futhark with generate code in the official Nim style
16:01:51FromDiscord<Chronos [She/Her]> Yeah fair
16:07:57*sagax quit (Read error: Connection reset by peer)
16:13:43*krux02 joined #nim
16:14:37FromDiscord<mratsim> In reply to @rubythulhu "and i wanted to": I never finished my 65c816 emulator:all those addressing modes .....↵https://github.com/mratsim/glyph/blob/master/glyph/snes/addressing_modes.nim
16:17:51FromDiscord<mratsim> sent a code paste, see https://play.nim-lang.org/#ix=html>
16:58:59FromDiscord<piqueiras> doing AoC today, anyone know how to do this in a cooler way? https://media.discordapp.net/attachments/371759389889003532/1186351918264631336/image.png?ex=6592ef52&is=65807a52&hm=e176382a71a6d765e09491690d1c9c205bc691136be19be5e09948a1fceea495&
16:59:10FromDiscord<piqueiras> (i am aware its not efficient)
17:02:22FromDiscord<metagngn> probably using strscans
17:02:27FromDiscord<metagngn> underused module
17:07:41FromDiscord<_goel_> Does nim sequences also have a `capacity`? Or just the length
17:07:47FromDiscord<_goel_> (edit) "Does" => "Do"
17:08:50FromDiscord<_goel_> (edit) "sequences" => "`seq`"
17:14:36FromDiscord<pmunch> In reply to @jviega "Yeah but I've written": Exactly, I've ended up using Futhark way more than I anticipated. But many of my projects ends up never being used by me.
17:15:30FromDiscord<pmunch> In reply to @_goel_ "Do nim `seq` also": It has a capacity internally. But it's not accessible by default
17:16:15FromDiscord<jviega> Well you can set it at the beginning but not query it
17:16:20FromDiscord<piqueiras> In reply to @metagngn "probably using strscans": hmm
17:16:21FromDiscord<jviega> newSeqOfCap()
17:26:07FromDiscord<0ffh> sent a code paste, see https://play.nim-lang.org/#ix=html>
17:29:07FromDiscord<0ffh> I can't even see any reason why it would want to evaluate "i" at compile time.
17:34:42*xet7 quit (Remote host closed the connection)
17:46:55*xet7 joined #nim
17:47:15*lumidify quit (Quit: leaving)
17:51:57*lumidify joined #nim
18:32:52FromDiscord<odexine> needs more context, it does indeed make no sense for that to fail though
18:34:44FromDiscord<graveflo> doesnt it look like the fist parameter to `[]` is defined as a `static` int?
18:39:49FromDiscord<odexine> only reason for that to be is if t is a tuple; additionally the other two wouldnt work either
18:40:02FromDiscord<bostonboston> Is there a way to do `recvFrom` from std/net with a timeout
18:45:20FromDiscord<0ffh> sent a code paste, see https://play.nim-lang.org/#ix=html>
18:45:53FromDiscord<odexine> tuples require compile time values for indexing
18:46:10FromDiscord<odexine> otherwise you have a runtime value that can change return types
18:47:19FromDiscord<odexine> i'd recommend going to an object in that case, that just looks like tuple abuse
18:47:28FromDiscord<0ffh> Meh, I need to store those contiguously in an array for performance reasons.
18:48:37FromDiscord<0ffh> then again, there's already lots of indirections in there
18:48:41FromDiscord<odexine> the issue iirc is basically youre conflicting with the generic tuple [] access
18:48:55FromDiscord<0ffh> ah!
18:49:00FromDiscord<odexine> basically just change the tuple into an object and id say its prolly gonna work
18:54:10FromDiscord<0ffh> Object works, yes.↵I wonder though, my index type is a seq[int] while I assume that the generic tuple [] takes an int.
18:54:47FromDiscord<0ffh> So should they really conflict?
18:56:11FromDiscord<graveflo> idk if this is the reason, but "array get" syntax has it's own semantics in the compiler. So do dot expressions for that matter, but that doesn't really mean that this behavior is expected. Yuo're using the tuple like an object, so it could be that there is something weird going on becuase that is not common
18:57:08FromDiscord<graveflo> by that I mean, the overload resolution might not even be happening the same way in the case for array get syntax
19:00:06FromDiscord<0ffh> Well, it seems like my hands are bound, then.↵Sooo many indirections, though! 😢
19:01:10FromDiscord<0ffh> This will have totally messed up memory access patterns.
19:14:59FromDiscord<graveflo> Yea, just for posterity it looks like the `[]` operator on tuple types is hard checked in the compiler. Overload resolution won't even happen if the offset can't the evaluated at compile time.
19:15:27FromDiscord<graveflo> (edit) "Yea, just for posterity it looks like the `[]` operator ... on" added "in (subscript notation)"
19:25:32FromDiscord<pmunch> In reply to @0ffh "Well, it seems like": Objects and tuples are laid out the same in memory AFAIK, so it won't change any indirections or memory access patterns to go from tuple to object
19:25:36FromDiscord<Chronos [She/Her]> https://github.com/Yu-Vitaqua-fer-Chronos/Zaphyt if anyone cares to look through this aha, anyone know why I could possibly be getting a SIGILL here? :/↵↵I have no clue, honestly
19:25:59FromDiscord<Chronos [She/Her]> Ah hold on-
19:26:54FromDiscord<Chronos [She/Her]> There
19:27:08FromDiscord<graveflo> In reply to @0ffh "Well, it seems like": there might be a weird work around for this, where if you add an extra parameter to `[]` overload resolution will happen and the compile time check is skipped. It's not a "solution" by any means because its so stupid, but I think it might actually work
19:27:45FromDiscord<Chronos [She/Her]> https://github.com/Yu-Vitaqua-fer-Chronos/Zaphyt/blob/main/parser/types.nim#L62 seems to be here, specifically
19:27:58FromDiscord<Chronos [She/Her]> For some reason the AST is nil...
19:28:50FromDiscord<Chronos [She/Her]> Maybe I'll restart from scratch, actually
19:28:53FromDiscord<0ffh> In reply to @pmunch "Objects and tuples are": Hmmm, I might have got that wrong.↵So, if I've a seq of tuples I imagine them being one after the other in memory.↵A seq of objects, I imagine a list of pointers.↵Is that wrong?
19:29:23FromDiscord<pmunch> Yes, that is wrong
19:29:35FromDiscord<graveflo> no the objects should all the have the same size so contiguous storage is still possible. If they are ref or ptrs then what you were saying is true
19:29:54FromDiscord<pmunch> If you have a sequence of `ref object` then they'd be pointers
19:30:03FromDiscord<griffith1deadly> In reply to @0ffh "Hmmm, I might have": seq of objects true to pointers only if object `ref object`
19:30:10FromDiscord<0ffh> Oops, sorry for the misunderstanding, then!↵Okay, objects are fine then!
19:30:50FromDiscord<0ffh> Still not quite used to the details of Nim's semantics.
19:32:27FromDiscord<0ffh> Or, semantics more or less, but not the implementation details.
19:38:36FromDiscord<JJ> you can think of objects as structs. and tuples as anonymous structs
19:39:12FromDiscord<JJ> despite the name, there's little object-oriented about them: all functions defined on them, even those called with `Object.function`, are statically dispatched
19:39:29FromDiscord<JJ> (you can write `method`s that are dynamically dispatched but there is little reason to do so)
19:40:21FromDiscord<0ffh> In reply to @omentic "despite the name, there's": Yeah, that part I know. It's the same UFCS as in D.
19:40:51FromDiscord<JJ> In reply to @omentic "you can think of": ah, i say this, but actually objects must have names for all of their fields: you can't treat a tuple as an object directly
19:40:54FromDiscord<griffith1deadly> In reply to @omentic "(you can write `method`s": what about inheritance?
19:41:47FromDiscord<JJ> inheritance is done with `object of RootObj` and `ref object of <name>`
19:42:41FromDiscord<JJ> ah, i see what you mean
19:42:48FromDiscord<bostonboston> Inheritance can be a valid use case of dynamic dispatching, but up can use the data of the object to determine calls in a lot of cases
19:43:29FromDiscord<griffith1deadly> sent a code paste, see https://play.nim-lang.org/#ix=html>
19:43:38FromDiscord<JJ> personally i shy away from inheritance in favour of variant objects whenever possible
19:43:40FromDiscord<bostonboston> The std/streams implementation for example
19:44:17NimEventerNew question by GiMS_Pang: Declare procedures prototypes in nim, see https://stackoverflow.com/questions/77681392/declare-procedures-prototypes-in-nim
19:46:55FromDiscord<griffith1deadly> In reply to @omentic "personally i shy away": you use it with object or ref object?
19:48:03FromDiscord<0ffh> In reply to @NimEventer "New question by GiMS_Pang:": Oof, needs forward declaration and will also explode the stack.
19:50:18FromDiscord<JJ> In reply to @griffith1deadly "you use it with": ref object only when i need indirection, for ex. an ast
19:51:32FromDiscord<JJ> but i'm not really defining different functions on different variants, i use it to define the same function, behaving differently
20:19:17*NimBot joined #nim
20:19:38*FromDiscord joined #nim
20:21:25FromDiscord<alec.j> How do I load an array from json? My json looks something like this: `{"arr": ["element", "element2", "element3"]}`
20:21:49FromDiscord<alec.j> I wanna get the array elements as a sequence preferably
20:26:42FromDiscord<Phil> In reply to @alec.j "I wanna get the": Have you checked out std/json already?↵Notably the `parseJson()` and `to` procs are of interest for you
20:27:04FromDiscord<alec.j> I found it lol
20:27:20FromDiscord<alec.j> Weird how it's called getElems and not getArray or smth
20:27:38FromDiscord<Phil> Generally I'd recommend bookmarking the overview of the stdlib 😉
20:28:16FromDiscord<Phil> In reply to @alec.j "Weird how it's called": It grabs you the individual elements as JsonNodes.↵To transform the JsonNode representing the entire list into a seq, use `to`
20:55:59FromDiscord<Phil> Question, what are the implications of using an atomic bool for a `while true` loop (that you can then kill multiple of by changing a single atomic bool).↵Does the bool being atomic make iterations "slower" in any meaningful capacity?
20:57:59FromDiscord<Elegantbeef> Depends on the cpu
20:58:01FromDiscord<Elegantbeef> But generally negligably
20:58:37FromDiscord<Phil> I'm mostly wondering if it would be a good or bad idea to have all my while-loops that you may spawn on n-threads depend on the same atomic.
20:58:41FromDiscord<Elegantbeef> Atomics have operations that are guaranteed to take one cycle, they do not use the same registers as non atomics so it can be slower, but it highly depends on the CPU
20:58:49FromDiscord<Elegantbeef> It's fine to do that
20:58:53FromDiscord<Phil> Like imagine 4 threads all on the same while-loop reading the same atomic until it gets switched
20:58:55FromDiscord<Phil> Check
20:59:35FromDiscord<Phil> Timeot implement a "killAll" proc on top of the procs that already exist to kill individual threads
20:59:43FromDiscord<Phil> threadservers
21:00:35FromDiscord<Elegantbeef> You probably do not even need an atomic here though, it saves one iteration at most
21:00:49FromDiscord<Elegantbeef> Just use it, but just saying
21:01:02FromDiscord<graveflo> is it imperative that the boolean is synchronized? If it's just a simple kill switch then you don't really have a benefit
21:01:32FromDiscord<Elegantbeef> Unless it is of utmost importance that the threads do not run another iteration there is really not much difference between the two
21:01:59FromDiscord<Phil> More a convenience utility. You currently can kill the threads individually but unless you raise an uncaught exception or defect in the main thread you can't really kill the program without doing so. A really easy graceful "Yeah just kill everything" would be neat
21:02:26FromDiscord<Phil> "Between the two"?↵What to? Did the bridge eat something?
21:02:36FromDiscord<graveflo> and thats assuming that the timing doesn't get shifted in the slightest and loop iterations are at the perfect ratio of long
21:03:08FromDiscord<Elegantbeef> The difference between global bool and a global atomic bool is that one will never run an extra iteration, whereas the other one might run another iteration
21:03:10FromDiscord<Phil> Or do you mean between the just "while true" that you can break out of with a Channel-message vs. changing an atomic
21:03:17FromDiscord<graveflo> just have a pointer to a bool maybe? flip the switch and they all die
21:03:59FromDiscord<graveflo> or that would just be an indirection. idk how this shit gets optimized. Maybe the global sticks in a register idk
21:04:15FromDiscord<Elegantbeef> Bleh it's generally the same
21:04:29FromDiscord<Phil> I think a global var bool should likely already do, which will just be a pointer underneath likely
21:05:48FromDiscord<Phil> or a ref
21:07:26FromDiscord<Elegantbeef> Using the atomic is fine
21:07:27FromDiscord<Phil> I mean if I can just do a global var bool that is easier and less likely to blow up in my face in case this ever starts mattering:↵> Types and operations for atomic operations and lockless algorithms.↵> ↵> Unstable API.
21:07:27FromDiscord<Phil> Less dependencies is always better
21:07:29FromDiscord<Elegantbeef> Well like I said with a non atomic there is the chance for one extra iteration
21:07:38FromDiscord<Phil> That shouldn't really matter I don't think.↵If it does, somebody will raise an issue for it.
21:07:46FromDiscord<Elegantbeef> I say one extra but I guess it's N extra
21:07:57FromDiscord<Phil> I highly doubt that would be the thing that makes somebody jump ship
21:08:04FromDiscord<graveflo> I would think of that more like an "iteration after" .. its different in such a small way but also ina big way
21:08:38FromDiscord<Elegantbeef> Yea if the thread is dealing with resources that the main thread has destructed it can be bad
21:09:11FromDiscord<Phil> But in those scenarios I feel like you'd be more trying to go for a controlled shut down
21:09:36FromDiscord<Phil> Which means you either shut down the threads yourself individually with already provided utilities or... can't think of anything else
21:12:46FromDiscord<rubythulhu> In reply to @mratsim "I never finished my": haha yeah there’s a lot of them. also to make the long addresses work it sends half the address on the rising edge of clock and the other half on the falling edge, which is a cute trick but makes the hardware impl talking to it more crazy
22:04:09FromDiscord<graveflo> trying to get nesper to work with atlas. I'm told to put an `include` statement in `X.nimble` so I can run some tasks, but I cannot figure out how to get nimble to run with the correct path configuration to find the nesper package installed in atlases custom nimble path. I know I can just run the configurations manually, but is there a way to set the nimble path used by nimble? I tried `nim.cfg` and `config.nims` and stuff like that and
22:04:33FromDiscord<graveflo> (edit) "atlases" => "atlas's"
22:08:42FromDiscord<Elegantbeef> Just use a relative path no?
22:08:53FromDiscord<Elegantbeef> `yourDeps/nesper/build_utils/tasks`?
22:09:25FromDiscord<graveflo> yea ig I could just do that. it would be more like `../deps/nesper/src/nesper/build_utils` blah
22:09:50FromDiscord<Elegantbeef> Well atlas does `--noNimblePath` so I assume that it makes nimble very silly on path resolution
22:09:52FromDiscord<graveflo> but it would be nice if I could tell nimble to use a custom nimble path for itself in this script
22:10:13FromDiscord<Elegantbeef> Really though you should just do a `myProject.nims` and use `nim task`
22:10:23FromDiscord<Elegantbeef> Unless nesper relies on nimble functionality
22:14:28FromDiscord<graveflo> unfortunately, it looks like it uses some nimble related macros, specifically `after` so idk if that makes it nimble dependent or not
22:23:40FromDiscord<Phil> hmmm I wonder if owlkettle has a graceful programmatic way to kill the GTK eventloop
22:25:50FromDiscord<Phil> It does! `g_main_loop_quit()`
22:41:55*advesperacit quit ()