<< 01-01-2021 >>

---

00:00:04	FromDiscord	<hugogranstrom> Happy new year 🎇🎆🥳
00:01:39	FromDiscord	<lqdev> happy new year!
00:02:01	*	JustASlacker quit (Ping timeout: 264 seconds)
00:02:11	FromDiscord	<lqdev> though it's already 1 am where i live ·–·
00:02:37	FromDiscord	<hugogranstrom> Same, I'm juste late hehe
00:03:21	FromDiscord	<lqdev> 2021 oughtta be an exciting year for nim
00:04:48	*	hmmm quit (Quit: WeeChat 3.0)
00:05:34	*	lritter quit (Quit: Leaving)
00:06:01	*	vicfred quit (Quit: Leaving)
00:06:13	FromDiscord	<shadow.> yeah for sure
00:11:13	FromDiscord	<William_CTO> Merry 2021
00:11:15	FromDiscord	<hugogranstrom> Absolutely, momentum is building up
00:12:14	FromDiscord	<bark> happy birthday to the people born on january the first!
00:14:36	FromDiscord	<arun> sent a long message, see http://ix.io/2Ksb
00:16:26	FromDiscord	<dom96> haha, what Araq said is now immortalised into a blog post
00:18:45	*	kenran quit (Quit: leaving)
00:19:16	*	vicfred joined #nim
00:25:03	*	NimBot joined #nim
00:41:01	*	tane quit (Quit: Leaving)
01:07:04	FromDiscord	<Vindaar> happy new year everyone 🥳
01:08:02	FromDiscord	<lqdev> ~~vindaar you're late~~
01:08:07	FromDiscord	<lqdev> er
01:08:10	FromDiscord	<lqdev> happy new year!
01:24:59	*	smitop quit (Quit: Connection closed for inactivity)
01:29:18	FromDiscord	<juliuskiesian> is there a quick way to convert a set to a seq?
01:29:34	FromDiscord	<Recruit_main707> lets go, we made it into the "name is even shown in the graph" (top 21) in most used languages of AoC https://media.discordapp.net/attachments/371759389889003532/794376739153379328/unknown.png
01:29:34	FromDiscord	<ElegantBeef> Yes in devel
01:30:12	FromDiscord	<ElegantBeef> https://nim-lang.github.io/Nim/setutils.html#toSet.t%2Cuntyped
01:30:17	FromDiscord	<ElegantBeef> Courtesy of me 😛
01:38:20	FromDiscord	<juliuskiesian> sent a code paste, see https://play.nim-lang.org/#ix=2Ksu
01:38:36	FromDiscord	<juliuskiesian> do i need latest devel?
01:39:10	*	Prestige quit (Quit: Prestige)
01:39:59	FromDiscord	<ElegantBeef> `std/setutils`
01:40:12	FromDiscord	<ElegantBeef> All new modules are forcing that `std`
01:40:32	*	Prestige joined #nim
01:40:41	FromDiscord	<ElegantBeef> It should be used regardless and i think should throw an error if not using `std` for any of the stdlib from your own project with a flag to disable that
01:52:28	Torro	happy new year
01:52:36	*	Torro quit (Quit: bye)
01:54:03	*	Prestige quit (Quit: Prestige)
01:55:23	*	Prestige joined #nim
02:24:29	*	disruptek throbs.
02:43:46	Prestige	disruptek: will you throb less in 2021?
02:48:48	disruptek	prestige: will you sponsor me?
02:52:45	Prestige	Like, AA?
02:53:00	Prestige	or like a commercial sponsor
02:53:22	disruptek	yes.
02:53:35	Prestige	Sure, if you bend to my will for a task
02:53:45	disruptek	i signed up for github sponsors to help support my drug habit.
02:54:03	disruptek	what's the task?
02:54:26	Prestige	multiple inheritance or concepts
02:54:36	disruptek	i fucking knew it.
02:54:40	FromDiscord	<ElegantBeef> Prestige i'm working on the `absImport` now, got the AST generated but it's a nice error generating macro
02:54:59	disruptek	what's concepts worth to you?
02:55:19	Prestige	in dollars? hmm
02:55:44	Prestige	oh beef, neat
02:56:05	disruptek	for a nim programmer, beef has very poor taste.
02:56:11	FromDiscord	<ElegantBeef> Truth
02:56:22	FromDiscord	<ElegantBeef> He's the one that suggested it so meh
02:56:33	FromDiscord	<ElegantBeef> What's the issue with my taste anywho?
02:57:02	disruptek	iirc, absImport is a bad idea.
02:57:21	FromDiscord	<ElegantBeef> Project root relative importing
02:57:44	FromDiscord	<ElegantBeef> Prestige suggested it so send the hate to him
02:57:50	Prestige	ye
02:57:53	disruptek	you say tomato, i say dogshit.
02:58:16	Prestige	eh I would like to have it
02:58:35	Prestige	anyway disruptek idk the current state of concepts
02:58:50	FromDiscord	<ElegantBeef> Nice error anywho, AST is identical to what a user would write but it's saying the file doesnt exist
02:58:52	FromDiscord	<ElegantBeef> So 10/10
02:59:13	disruptek	the current state is that i refuse to test them until the old syntax is ported, and araq says that i'm an idiot as a result.
02:59:27	disruptek	actually, he probably has other reasons for calling me an idiot.
02:59:53	FromDiscord	<ElegantBeef> I'm surprised i havent been called an idiot by him yet 😛
03:00:21	disruptek	but the fact is, i don't see how we can merge them without a migration.
03:01:19	disruptek	old concepts are not just slow, but very broken.
03:01:29	disruptek	and recently, they even broke jason.
03:01:37	disruptek	and my json serializer, also.
03:02:28	Prestige	that's a sad state of affairs
03:02:36	disruptek	it's not brilliant.
03:16:35	FromDiscord	<ElegantBeef> So prestige doesnt seem there is an overly pretty way of doing it
03:16:49	FromDiscord	<ElegantBeef> I've got a template call which works properly but then the import fails
03:17:29	FromDiscord	<ElegantBeef> It's pushed if interested in lookging
03:18:53	Prestige	Link?
03:19:49	FromDiscord	<ElegantBeef> https://github.com/beef331/import-utils
03:25:48	disruptek	Link?
03:26:41	disruptek	did you really make a repo with a fucking hyphen in it?
03:27:05	disruptek	what did i do to you?
03:29:11	Prestige	@ElegantBeef are you just getting the error in the tests that submod doesn't exist?
03:29:23	disruptek	i need to add a package section to my style doc.
03:29:44	FromDiscord	<ElegantBeef> Thought i made it with an underscore to be honest
03:29:47	FromDiscord	<ElegantBeef> atleast i intended to
03:30:31	Prestige	disruptek: How much for a proper multiple inheritance implementation? I know how it should be structured, but I'm not a nim wizard. And it would require compiler changes afaict
03:31:18	disruptek	you /do/ know how it should be structured?
03:31:29	Prestige	Yes
03:31:39	disruptek	!rfc inheritance
03:31:39	disbot	https://github.com/nim-lang/RFCs/issues/262 -- 3Support for Multiple Inheritence 7& 7 more...
03:31:53	Prestige	Damn that bot is handy
03:32:18	Prestige	I'd think the implementation shown here would suffice https://oberoncore.ru/_media/library/templ_a_systematic_approach_to_multiple_inheritance_implementation.pdf
03:32:55	Prestige	I just have one particular point that should be added, which is illustrated in my RFC. Might not be super clear
03:33:14	disruptek	"none of these arguments hold." -- so, it doesn't increase expressive power, either?
03:33:34	disruptek	which language is your MI experience with?
03:33:36	disruptek	java?
03:33:40	Prestige	Idk what he means by that
03:33:56	Prestige	eh java only has a half baked implementation of MI
03:34:10	disruptek	don't make me guess.
03:34:12	Prestige	It's just a subject I've thought about and have wanted
03:34:13	disruptek	python?
03:34:15	Prestige	nah
03:34:24	Prestige	I've not used it in other languages
03:34:35	Prestige	Just have wished I'd had it available
03:34:46	*	vicfred quit (Quit: Leaving)
03:35:09	Prestige	mainly in java and typescript (and nim, now)
03:36:05	disruptek	hmm.
03:36:07	Prestige	I've mainly wanted it for game development
03:37:05	disruptek	why do you think it needs compiler changes?
03:37:14	Prestige	Anyway this is my main point here https://github.com/nim-lang/RFCs/issues/262#issuecomment-713236634
03:37:15	disbot	➥ Support for Multiple Inheritence ; snippet at 12https://play.nim-lang.org/#ix=2Jbu
03:37:16	disruptek	i will read the oberon paper, but first i want to pick your brain.
03:38:03	disruptek	"Both arguments are false as will be shown below."
03:38:16	disruptek	so, yeah, it doesn't increase expressive power.
03:38:16	Prestige	Basically new procs (or methods?) need to be created for the sub-type, which may need to remove a procCall to a grandparent so that it is not invoked twice
03:38:46	disruptek	no big deal.
03:39:03	Prestige	I think it does, but I didn't understand why he said it wouldn't
03:39:16	disruptek	what would be perhaps more valuable would be a proper vtable impl.
03:40:02	disruptek	i'm a little sad that we gave up multi-methods; how complex could they be, right?
03:40:07	Prestige	I think he goes into that about a page or two in, for optimisations
03:40:28	disruptek	it matters to us because we do method dispatch using method /names/ at the moment.
03:40:37	Prestige	Ah, I see
03:41:01	disruptek	this hampers adoption of inheritance, which hampers interest and performance of MI.
03:41:39	Prestige	Yeah, I think we'd want vtables for this MI implementation as well, but you'd know better than I
03:41:46	Prestige	especially after reading the paper
03:43:29	Prestige	It's an interesting approach. I think I'll write up an actual MI use case for something I wanted to make
03:43:56	disruptek	that would be useful.
03:44:23	disruptek	what has discouraged you from building this?
03:45:32	Prestige	I'm not sure how it could be done outside of compiler chanegs, and I'm not sure where to look. It seems daunting, because my Nim knowledge is lacking
03:46:53	disruptek	in my limited experience, the compiler is the thorniest nim code you're likely to read.
03:47:15	disruptek	i'd much rather start from scratch.
03:47:50	Prestige	Hopefully it isn't too bad to deal with
03:48:19	disruptek	the description is three pages long. how hard could it be?
03:49:20	Prestige	¯\_(ツ)_/¯
03:51:07	disruptek	pick a syntax that we can impl, for starters.
03:51:37	*	muffindrake quit (Ping timeout: 260 seconds)
03:51:51	Prestige	I was thinking of using what was in my RFC, using commas to separate the types. Like `Player = ref object of Locatable, Updatable`
03:52:39	disruptek	type Missile = ref object {.inherits: [Projectile, ParticleEmitter].} might work for the macro impl.
03:53:10	disruptek	i don't think it makes sense to wrap the whole type section, and you cannot pragma type sections iirc.
03:53:11	Prestige	That also works, I just thought the former was a bit easier on the eyes
03:53:18	Prestige	true
03:53:23	disruptek	this is just the prototype, chucklehead.
03:53:38	Prestige	Whatever floats your boat, I'm not too picky
03:53:46	*	muffindrake joined #nim
03:57:40	saem	I also find it hilarious of using multiple inheritance in an example which so clearly demonstrates why the games industry went towards things like ECS (composition).
03:57:45	disruptek	if we can use something better than multiple dispatch I'm all for it. My main goal is to enable the programmer to have the functionality/syntax as I've described in the post.
03:58:12	disruptek	what /is/ that functionality?
03:58:12	Prestige	https://play.nim-lang.org/#ix=2KsV disruptek something like this
03:58:37	disruptek	it's like you want the programmer to have to write code and then the MI scheme deletes it.
03:58:48	disruptek	isn't this is opposite of what everyone else wants?
03:59:03	Prestige	I need to add something actually
03:59:07	Prestige	And no, not really
04:01:42	disruptek	procCall Entity(this).update(deltaTime)
04:01:47	Prestige	Here we go: https://play.nim-lang.org/#ix=2KsZ
04:01:49	disruptek	is this really what you want to type?
04:02:10	disruptek	why not `this.entity.update deltaTime`?
04:02:15	saem	Engines are running away from exactly this type of code.
04:02:42	Prestige	disruptek: I didn't know you could do that (unless you mean the new implementation)
04:03:26	disruptek	your player can have a field that holds the Entity in it, of course.
04:04:01	Prestige	Yeah, in accordance to the implementation in the paper, that's what would happen pretty much
04:04:07	saem	But why bother decomposing a player into components if you're just going to statically (MI) inherit it again?
04:04:13	disruptek	that's my point.
04:04:55	Prestige	What do you mean?
04:04:59	saem	I've done enough ECS like things, heck anyone has if they've used a database without some over wrought ORM, and this is exactly in the wrong direction.
04:05:25	saem	`type Player = ref object {.inherits: [Animation, Locatable].}` <-- this line is static, it's not runtime, happens once.
04:05:31	Prestige	I have yet to see an ECS implementation that wasn't terrible to work with
04:05:45	disruptek	!repos ecs
04:05:46	disbot	https://github.com/yglukhov/ecs -- 9ecs: 11Entity Component System 15 21⭐ 2🍴
04:05:46	disbot	https://github.com/rlipsc/polymorph -- 9polymorph: 11Polymorph is a composable entity component system generator - create statically defined ECS architectures generated from your design. 15 12⭐ 0🍴
04:05:46	disbot	https://github.com/planetis-m/breakout -- 9breakout: 11Game implemented using strict ECS architecture. Used as a testbed. 15 8⭐ 0🍴 7& 5 more...
04:06:12	disruptek	you looked at the 8 hits for `ecs` in nim?
04:06:31	Prestige	Yeah, I've looked at a few at least
04:06:55	disruptek	and what do they get wrong?
04:07:08	disruptek	besides lack of MI.
04:07:26	Prestige	They seem cumbersome to work with, it's a pain to use
04:07:31	saem	The querying is often not great in ECS implementations and that's where the annoying ceremony comes in most often.
04:08:10	saem	I think spending some time on a query system would actually make ECS quite pleasant. FlatDB actually has a pretty nice example of how to make a small and pleasant DSL for querying.
04:08:13	saem	in Nim.
04:08:24	disruptek	!repo flatdb
04:08:25	disbot	https://github.com/enthus1ast/flatdb -- 9flatdb: 11small/tiny, flatfile, jsonl based, inprogress database for nim 15 29⭐ 1🍴
04:10:01	Prestige	basically I'd want to do something like `type Player = ref object Animation, Locatable`, idk how that differs from the inherits pragma
04:10:35	disruptek	it differs because .inherits can be impl as a macro, today, without compiler changes.
04:10:43	Prestige	ah
04:11:02	saem	I don't believe you do, but who knows. I think what you want is the animate system, and it should see all entities with animation and location components and you want a nice way to receive that as a parameter and then do your thing.
04:11:15	Prestige	Needing the contents of other methos seems difficult though, I'm not sure how you do that with a macro
04:12:08	Prestige	saem: if you can make an ecs system that isn't annoying to use I'm all for it. I just have yet to see one
04:12:09	disruptek	you do that part with term-rewriting, probably, but i'm less and less convinced this is even worth building.
04:12:11	saem	so proc animation(a: AnimationTypeWhichHidesTheBoilerPlate): void
04:12:59	saem	AnimationTypeWhichHidesTheBoilerPlate: is an entity with Animation and Locatable components.
04:13:00	Prestige	disruptek: what are you not convinced of?
04:13:24	disruptek	that this is even worth building, of course.
04:13:34	Prestige	I mean, why?
04:13:43	disruptek	because i read the paper.
04:14:13	Prestige	Can you elaborate why it convinced you it isn't worth building?
04:14:21	disruptek	the syntax looks poor, the speed is worse due to the compiler impl, and it doesn't buy you anything in terms of expressive power.
04:14:35	disruptek	what have i missed?
04:14:47	disruptek	it's a bunch of extra code that you don't need.
04:14:55	disruptek	it buys you nothing and can only limit you.
04:15:15	Prestige	It gives more power to the programmer to be able to use MI. MI wasn't implemented in several languages for no reason; it's useful
04:15:28	disruptek	uh, that's not what the paper says.
04:15:38	disruptek	do you wonder why it wasn't impl in nim?
04:15:51	Prestige	There are many things not implemented in Nim
04:16:07	FromDiscord	<Jarred Kenny> 🍿
04:17:57	Prestige	The only way to implement the example architecture without MI (with many more objects and types) is to create an elaborate ECS system, that will generate much more code and will be cumbersome to use in comparison to multiple inheritance
04:19:31	disruptek	ask lqdev what he thinks.
04:20:32	Prestige	of what, a 3rd way to implement the architecture?
04:21:01	disruptek	no, go like, "lqdev: am i a knucklehead?"
04:21:05	disruptek	and see what he says.
04:21:24	saem	https://play.nim-lang.org/#ix=2Kt3
04:22:14	Prestige	saem: yeah, I've seen ECS stuff before
04:22:18	saem	That's probably closer to what you want. The you can write proc on the type that is the query, so it looks like a type that's all once pretty object, but not all the rest of what MI entails which is layout of memory and all sorts of other bits.
04:23:58	saem	data oriented stuff is definitely weird because it's exactly not domain oriented. It's really annoying when all you want to do is reason about a player and things they do. Not worry about first decomposing that into all the systems, components, etc...
04:25:45	saem	I don't think MI helps there, unfortunately. But I do get your point about wanting a layer on top and not have to hand roll all that decomposition yourself and how it might seem like the MI approach might do that.
04:26:08	disruptek	i don't think it helps at all.
04:26:38	saem	Mechanically no, but I get why ones intuition might drive in that direction.
04:26:56	disruptek	too much drinking in college?
04:26:59	Prestige	It does all the composition in a simple way with almost 0 extra code, without an external framework, creating a bunch of systems etc
04:27:21	disruptek	there's plenty of extra code.
04:27:25	Prestige	I would like to use ECS, as I've been saying for like 8 months, but I just find all the implementations lacking
04:27:27	disruptek	good lucky optimizing around it.
04:28:04	Prestige	Nothing needs to be optimized. And no extra code for the one using MI
04:28:09	disruptek	polymorph seems to have a lot of... reasoning.
04:29:00	disruptek	did you try to build anything with it?
04:29:01	saem	The thing with ECS is that it's meant to give you benefits around memory layout and the ability to dynamically change behaviour by adding and removing components. MI ties your hands because it forces specific memory layouts.
04:29:16	Prestige	disruptek: with ecs? Yeah
04:29:16	disruptek	that's my point.
04:29:31	disruptek	with polymorph?
04:29:39	Prestige	saem: yes it does, but I don't care about that
04:29:56	Prestige	I'd rather just be able to write the code I want without micro optimising
04:30:29	disruptek	i think you would be better served by some templates.
04:30:32	Prestige	If I were writing a game engine for commercial games or something, I might want ECs
04:31:16	Prestige	disruptek: you mean making something with inheritance? I have, just not MI (because it doesn't exist)
04:31:19	saem	OK, so that's all statically determined, in which case can't you use concepts?
04:31:29	disruptek	i mean making something with polymorph.
04:31:41	disruptek	for the third time...
04:32:02	Prestige	Afaik concepts aren't working entirely, and they only serve part of the functionality that MI would
04:32:34	Prestige	disruptek: idk what you're asking. Is that a library or something?
04:32:49	*	disruptek 🤦
04:33:16	disruptek	i want you to look at these two links:
04:33:19	disruptek	!repo pixecs
04:33:20	disbot	https://github.com/PixeyeHQ/pixecs -- 9pixecs: 11A pragmatic entity-component-system module for my gamedev needs 15 22⭐ 0🍴
04:33:22	disruptek	!repo polymorph
04:33:23	disbot	https://github.com/rlipsc/polymorph -- 9polymorph: 11Polymorph is a composable entity component system generator - create statically defined ECS architectures generated from your design. 15 12⭐ 0🍴 7& 1 more...
04:34:40	Prestige	I believe I looked at polymorph a few months ago
04:35:05	*	spiderstew joined #nim
04:36:01	*	spiderstew_ quit (Ping timeout: 246 seconds)
04:38:01	FromDiscord	<m0nsta> HNY Folks!
04:38:16	Prestige	Happy new year!
04:39:37	Prestige	well pixecs is out at least
04:44:08	Prestige	I think having a callback system with polymorph for object events would also be limited or not possible, because the object would be missing its type info...
05:05:57	*	vicfred joined #nim
06:06:35	FromGitter	<ornamentist> Can anyone point to a library with idiomatic examples of using Concepts?
06:06:57	disruptek	!repo jason
06:06:58	disbot	https://github.com/disruptek/jason -- 9jason: 11JSON done right 🤦 15 38⭐ 1🍴
06:07:58	disruptek	concepts are being reimpl'd, though. improved. be prepared to change your syntax somewhat.
06:08:02	FromGitter	<ornamentist> Excellent thanks
06:09:05	FromGitter	<ornamentist> Are there any notes I can read on the latest thinking on Concepts?
06:10:02	*	waleee-cl quit (Quit: Connection closed for inactivity)
06:15:13	FromDiscord	<Rika> prolly somewhere in the issues of the rfcs repo of nim
06:15:23	disruptek	!rfc concepts
06:15:25	disbot	https://github.com/nim-lang/RFCs/issues/168 -- 3Concepts and type-checking generics 7& 29 more...
06:15:35	FromDiscord	<Rika> oh you made an rfc command? nice
06:15:49	disruptek	it has always existed.
06:15:59	FromDiscord	<Rika> didnt know xd
06:16:08	disruptek	discord chumps.
06:16:13	FromDiscord	<Rika> lmao
07:02:30	*	Prestige quit (Quit: Prestige)
07:03:48	*	Prestige joined #nim
07:09:07	FromDiscord	<ElegantBeef> Lol forum doesnt do well with the new year all activity is now dec 2020 😄
07:12:30	*	Prestige quit (Quit: Prestige)
07:14:12	*	Prestige joined #nim
07:28:46	FromDiscord	<Rika> lmao
07:33:11	FromDiscord	<lantos> happy new year
07:47:14	*	narimiran joined #nim
07:55:21	*	habamax joined #nim
08:06:13	saem	Rolled around to my part of the world.
08:10:40	*	Prestige quit (Quit: Prestige)
08:11:55	*	Prestige joined #nim
08:17:59	FromGitter	<ornamentist> Is the convention in Nim projects to have a <dir>/<projectname> top-level directory for source files or <dir>/src?
08:19:19	FromDiscord	<ElegantBeef> the latter `nimble init` will set it up for you 😄
08:54:35	FromDiscord	<treeform> Ornamentist, I set up this template to setup new projects: https://github.com/treeform/nimtemplate
08:55:27	*	actuallybatman joined #nim
08:55:37	*	actuallybatman quit (Quit: Lost terminal)
08:56:31	saem	impossible, can't be batman, saw them pop in.
09:20:17	*	gangstacat quit (Quit: Ĝis!)
09:31:13	FromDiscord	<lantos> which nim vscode extention should we be using?
09:35:05	FromDiscord	<ElegantBeef> I'm trying saem's as he's improving it to be more stable/usable
09:39:03	FromDiscord	<flywind> I disabled `nimsuggest`, so I don't care which one to use 😜
09:39:22	FromDiscord	<ElegantBeef> Well saem's doing some good work, goto defs now work on the compiler
09:40:18	FromDiscord	<flywind> Does it need `nimsuggest`?
09:41:02	FromDiscord	<ElegantBeef> Believe so
09:41:50	FromDiscord	<flywind> Then I would rather use `seacrh` instead.
09:41:57	FromDiscord	<ElegantBeef> Why?
09:43:51	saem	I mean, I might be able to get nimfind in there soon enough.
09:44:19	FromDiscord	<flywind> `nimsuggest` always uses huge memory and even I close the vscode, they still occupy nim.exe.
09:44:48	saem	This is primarily for the compiler code base?
09:46:13	saem	I had to rework the whole elrpc stuff a bunch, part of that was better process clean-up IIRC.
09:47:12	saem	As someone who was starting up the extension and killing it like 30+ times a day that got on my nerves fast. I've encountered it once since then -- not perfect.
09:48:07	*	superbia joined #nim
09:52:34	FromDiscord	<ElegantBeef> Yea i mean saem's resolving these issues that are caused by spawning them up so much
09:52:48	*	gangstacat joined #nim
09:52:56	FromDiscord	<flywind> I see
09:53:25	saem	I still need to fix the whole default is one nimsuggest per file, but I have a working config for the compiler for that -- just need to auto-detect it and make the spawning lazy.
10:06:31	Prestige	Interesting discussion on the Nim 2020 HN post about debugging with gdb
10:06:43	saem	Oh yeah?
10:06:49	Zevv	yeah
10:07:05	Prestige	https://news.ycombinator.com/item?id=25590962
10:07:10	saem	thanks
10:07:35	Zevv	The last few months Nim HN posts did well, comment-wise
10:08:19	FromDiscord	<lantos> nim has a step through break point vscode plugin
10:08:57	Prestige	I think it's mentioned, but people said it would bug out sometimes
10:08:58	FromDiscord	<lantos> https://github.com/jasonprogrammer/nim-debug-example
10:10:09	saem	I stumbled upon that repo a while back but forgot about it
10:10:25	FromDiscord	<lantos> https://media.discordapp.net/attachments/371759389889003532/794507816123236363/unknown.png
10:11:31	saem	I just updated nim-gdb.py in the latest so it's a bit better, hopefully -- still plenty of gaps.
10:11:56	saem	choosenim ships with a broken nim-gdb and I haven't gotten around to submitting a PR to fix it.
10:12:29	FromDiscord	<lantos> could it be included in a vcode ext so that new people have it from get go
10:12:49	saem	That's a bit trickier than I would like, but yeah.
10:13:40	*	natrys joined #nim
10:13:58	saem	That's the intention but I'd like some more things cleaned up before hand. Otherwise it's a big time sink for a really mediocre version.
10:26:32	*	hnOsmium0001 quit (Quit: Connection closed for inactivity)
10:33:54	*	superbia left #nim ("WeeChat 3.0")
10:37:57	Prestige	I have a string I need to pass to a proc (c binding) which takes a Pcuchar - I've cast it, and it works fine. When I try to compile with orc, it says the expression cannot be cast to Pcuchar=cstring. Am I misunderstanding something?
10:38:34	*	MyMind quit (Quit: WeeChat 3.0)
10:58:22	FromDiscord	<Meowz> whats the name of nims code formatter again
10:59:42	FromDiscord	<Meowz> nimpretty, got it
11:22:59	Zevv	Prestige: got a code snippet?
11:28:26	*	JustASlacker joined #nim
11:58:22	FromDiscord	<juliuskiesian> is there a way to write something like `for x in 10 ..< 1`?
11:59:38	ForumUpdaterBot	New thread by Nobody: Arraymancer `<https://github.com/mratsim/Arraymancer>`, see https://forum.nim-lang.org/t/7318
12:00:22	FromDiscord	<flywind> ! eval for i in countdown(3, 1): echo i
12:00:29	FromDiscord	<flywind> (edit) "! eval" => "!eval"
12:00:53	FromDiscord	<flywind> !eval for i in countdown(3, 1): echo i
12:00:55	NimBot	3↵2↵1
12:01:28	Zevv	clyybber here?
12:01:40	FromDiscord	<juliuskiesian> ideally, 1 shouldn't be included. @flywind
12:02:29	FromDiscord	<flywind> you could always use `(i + 10, i + 1)`.
12:03:04	FromDiscord	<flywind> or you could your own function
12:05:29	*	tane joined #nim
12:07:04	FromDiscord	<flywind> see https://github.com/Yardanico/nimpylib/blob/adec704b91a66669686bf3f251320ba1f5d9a6cd/src/pylib/xrange.nim#L14
12:22:40	*	xet7 quit (Remote host closed the connection)
12:26:39	*	Arrrrrrrr joined #nim
12:30:35	Arrrrrrrr	Happy 2021, hope nim grows 2000%
12:35:23	FromDiscord	<Rika> Why not 2020%
12:50:51	Zevv	2021%, right
13:04:11	FromDiscord	<Sorrow> Love it so far. Shame there's no music tagging lib like Mutagen.
13:12:02	*	Vladar joined #nim
13:12:46	FromDiscord	<juliuskiesian> is there a quick way to convert a `Tensor[float]` from arraymancer to a seq?
13:14:16	FromDiscord	<Clyybber> Zevv: here
13:16:02	*	habamax quit (Quit: leaving)
13:16:17	FromDiscord	<juliuskiesian> bit convoluted, but i found a way: `toOpenArray(xs.dataArray, 0, xs.size - 1).toSeq`
13:16:58	Zevv	oi; you know some stuff about repr, reprv2, reprjs and dollars, right?
13:17:37	FromDiscord	<Clyybber> Zevv: just saw your pr, yeah
13:18:14	Zevv	Why is it that $ and repr don't use the same underlying implementations?
13:18:23	FromDiscord	<Ben Boughton> Hello! Brand new to nim. Quick question. How do I get a variable value during compilation?
13:18:38	FromDiscord	<lqdev> you can't
13:18:58	FromDiscord	<lqdev> elaborate on what you're trying to do
13:20:15	FromDiscord	<Ben Boughton> C:\Apps\nim-1.4.2\lib\system\fatal.nim(49, 5) Error: unhandled exception: C:\Users\bboug\.nimble\pkgs\nimterop-0.6.13\nimterop\cimport.nim(12, 14) `(not fail)` File or directory not found↵↵I want to echo the srcDir during compilation so I know exactly where it is looking
13:21:06	FromDiscord	<Clyybber> Zevv: In general or for sets?
13:21:48	Zevv	In general
13:21:53	FromDiscord	<lqdev> sent a code paste, see https://play.nim-lang.org/#ix=2KuM
13:22:07	FromDiscord	<lqdev> but you need to make sure that `yourSrcDir` is either a const or a {.compileTime.} var or let
13:23:26	FromDiscord	<Clyybber> Zevv: repr is mainly for debugging so it mught include stuff that you wouldn't want in `$` like the address a seq's data points to and stuff like that
13:23:43	Zevv	hm fair enough
13:23:56	Zevv	ok, so I need to figure out how to get my js repr to work from this funny set representation
13:24:01	Zevv	thanks
13:24:13	FromDiscord	<Ben Boughton> Fantastic! That works
13:24:55	FromDiscord	<Ben Boughton> I appreciate your help.
13:28:24	FromDiscord	<lqdev> np
13:45:33	*	vsantana joined #nim
13:45:48	*	waleee-cl joined #nim
13:50:16	*	Arrrrrrrr quit (Quit: Arrrrrrrr)
13:51:23	FromDiscord	<dom96> Zevv: repr is built to work on anything you throw at it, `$` needs to be defined for each type
13:52:12	*	superbia joined #nim
13:57:54	*	JustASlacker quit (Ping timeout: 256 seconds)
13:58:19	*	superbia quit (Quit: WeeChat 3.0)
13:58:21	*	hmmm joined #nim
13:59:40	Zevv	dom96: yeah, sure but for types where `$` is defined, repr could use `$` or the other way around. Now there are duplicate implementations.
13:59:57	ForumUpdaterBot	New thread by Alexeypetrushin: Seems like a bug? SIGSEGV: Illegal storage access., see https://forum.nim-lang.org/t/7319
14:00:41	FromDiscord	<dom96> `repr` gives a much more useful representation of data structures though
14:00:51	FromDiscord	<dom96> I don't want a string representation of JSON for example
14:00:57	FromDiscord	<dom96> I want to see what the data structure looks like
14:01:24	FromDiscord	<dom96> The same applies to strings if I want to see the escaped character literals instead of the characters they represent
14:01:42	Zevv	fair enough. I was adding a special case for `set[char]` that uses `..` instead of dumping all the individual chars, but I need to put my implementation in 4 different places now, all slightly different
14:08:13	FromDiscord	<Clyybber> Zevv: Btw do you know more details on https://github.com/disruptek/cps/issues/47?
14:08:14	disbot	➥ mutable params and the developers who love them
14:08:37	Zevv	I don't think its realistic
14:09:15	Zevv	When passing var args to a proc, you assum the proc will run as a whole before returing, so the callers original var arg is guarenteed to be always there during the execution of the function
14:09:38	Zevv	when something is suspendable, I don't think we can have that guarentee in any way, so imho we always copy or move the arg
14:09:48	Zevv	and just refuse to CPS procs with var args
14:09:55	Zevv	but of course mr stubborn thinks otherwise
14:11:39	Zevv	so you'd have to wait for him to wake from his slumbers to get more insight in his inner thoughts and feelings on this subject
14:11:53	Zevv	oi disruptek, wakey wakey
14:11:55	FromDiscord	<Clyybber> aight, disruptek
14:11:57	Zevv	that's enough
14:12:02	FromDiscord	<Clyybber> heh
14:12:15	Zevv	it's past nine over there
14:13:46	FromDiscord	<Clyybber> Zevv If we store the arg somewhere we can make it always be there for the duration of execution no?
14:14:04	Zevv	yeah but it's all about making it var
14:14:12	Zevv	so you should be able to mutate the callers original value
14:14:15	Zevv	which might not even exist anymore
14:14:34	Zevv	is a pointer that might point to nowhere
14:15:01	FromDiscord	<Clyybber> because it has been moved?
14:15:17	Zevv	or just because it went out of scope and no longer exists
14:15:21	FromDiscord	<Clyybber> I see
14:15:41	Zevv	you have proc a with a local i, which a calls proc b passing the local as a var arg. b suspends and a returns
14:15:44	Zevv	where does i go?
14:15:47	Zevv	*poof* gone
14:16:09	Zevv	so I say we just ignore this case for now. If you want to mutate something, use a ref
14:16:52	Zevv	I think the js backend smells funny
14:22:49	FromDiscord	<jseb> speaking of javascript…
14:25:27	FromDiscord	<jseb> sent a code paste, see https://play.nim-lang.org/#ix=2Kv9
14:28:04	*	Vladar quit (Quit: Leaving)
14:30:25	FromDiscord	<jseb> ah , looking at the generated javascript, i have the name hardcoded at generation. That's not what i wanted, but i understand (the macro system replaces immediately with the value of the AST name of filename variable)
14:36:28	FromDiscord	<flywind> you could use template to workaround this issue
14:36:32	FromDiscord	<flywind> (edit) "template" => "templates"
14:38:15	FromDiscord	<flywind> sent a code paste, see https://play.nim-lang.org/#ix=2Kvg
14:41:02	FromDiscord	<flywind> Yeah, you could use `static string`
14:41:14	FromDiscord	<flywind> sent a code paste, see https://play.nim-lang.org/#ix=2Kvh
14:46:50	FromDiscord	<jseb> sent a code paste, see https://play.nim-lang.org/#ix=2Kvm
14:48:15	FromDiscord	<lqdev> sent a code paste, see https://play.nim-lang.org/#ix=2Kvl
14:48:44	FromGitter	<offbeat-stuff> Hi
14:48:46	FromDiscord	<lqdev> (edit) "https://play.nim-lang.org/#ix=2Kvl" => "https://play.nim-lang.org/#ix=2Kvo"
14:49:35	FromDiscord	<lqdev> hi
14:49:38	FromGitter	<offbeat-stuff> How to autogen nim bindings for c++ wrappers
14:49:40	FromGitter	<offbeat-stuff> Like there are pkg-config --cflags and pkg-config --libs for the wrapper
14:51:51	FromDiscord	<haxscramper> I suppose you might run shell command at compile-time and do something with it, like filling values of `{.passl: "<some text>".}`
14:51:55	FromDiscord	<jseb> @lqdev : no, it stills resolves filename at the compilation time
14:52:24	FromDiscord	<lqdev> yeah but it will print the variable containing the filename
14:52:39	FromDiscord	<lqdev> why not just use `echo` or something instead of emitting a console.log?
14:52:50	FromDiscord	<Sorrow> sent a code paste, see https://play.nim-lang.org/#ix=2Kvp
14:53:13	FromDiscord	<lqdev> @Sorrow it's possible that the bodyStream contains less than 4194304 bytes.
14:53:26	FromDiscord	<Sorrow> The file's 1GB.
14:53:41	FromDiscord	<lqdev> how do you know it loads the whole file?
14:53:50	FromDiscord	<Sorrow> By monitoring it.
14:54:01	FromDiscord	<lqdev> monitoring what? `chunk`'s `len`?
14:54:06	FromDiscord	<jseb> yes you're right lqdev, `echo` is the way to go for this one. It was for testing `emit` macro
14:54:10	FromDiscord	<flywind> you could import `jsconsole` to use `console.log`.
14:54:16	FromDiscord	<Sorrow> Memory usage. It keeps increasing.
14:54:31	FromDiscord	<lqdev> the GC doesn't run immediately
14:54:54	FromDiscord	<lqdev> you can force a collection using `GC_fullCollect()`
14:55:12	FromDiscord	<lqdev> or use ARC or ORC which will free the data as soon as it's no longer in use
14:55:57	FromDiscord	<haxscramper> Or move `chunk` out of the loop and use it as buffer
14:56:04	FromDiscord	<haxscramper> WIth something like https://nim-lang.org/docs/streams.html#readData%2CStream%2Cpointer%2Cint
14:56:11	FromDiscord	<Sorrow> Thank you. I was thinking about using a buf instead.
15:15:21	FromDiscord	<bark> in nim, are ref objects required for dynamic dispatch?
15:17:00	FromDiscord	<haxscramper> IIRC yes
15:21:02	mipri	you can avoid ref objects, but you'll run into issues as soon as you want e.g. a seq[SuperClass]
15:23:44	mipri	https://play.nim-lang.org/#ix=2KvI this works f.e., but you can't ss.add those Shapes because they have different sizes
15:24:01	*	Vladar joined #nim
15:24:30	mipri	https://play.nim-lang.org/#ix=2KvJ fixing the warning
15:26:06	mipri	well size aside there's an ObjectAssignmentDefect
15:28:13	FromDiscord	<bark> https://play.nim-lang.org/#ix=2KvK has no ObjectAssignmentDefect
15:28:43	mipri	ss.add them instead of showId'ing them
15:29:45	FromDiscord	<bark> i guess its due to sizes
15:30:01	FromDiscord	<bark> i guess it's generally a good practice to make all inheritance chains ref
15:30:06	FromDiscord	<bark> or ptr
15:30:09	mipri	even when .sizeof is the same you get that error
15:30:23	FromDiscord	<bark> huh
15:30:51	FromDiscord	<Sorrow> How can I get the resp headers, status code etc before loading the file? Am using .get().
15:50:03	FromDiscord	<For Your Health> This is probably a dumb question, but what type of floats should I be using by default? I see some libs using float32, some using float64, and some using float.
15:52:54	FromDiscord	<InventorMatt> If you use float it should then default to either float32 or float64 based on what OS you are using for you
15:53:31	FromDiscord	<juan_carlos> architecture, not OS, 32 for 32bit, 64 for 64bit.
15:54:03	FromDiscord	<For Your Health> So is float generally preferred to use?
15:54:38	FromDiscord	<Anuke> According to this tutorial, `float` is always float64: https://nim-lang.org/docs/tut1.html#basic-types-floats
15:56:31	*	narimiran quit (Remote host closed the connection)
15:56:48	Zevv	use float unless you have reasons not to
15:57:01	Zevv	it's a cdouble under the hood indeed
15:58:24	FromDiscord	<juan_carlos> "Everything floats down here"
15:59:10	FromDiscord	<For Your Health> Sounds good. Thanks for the help
15:59:31	FromDiscord	<flywind> How can I `hash` float/Number in `Javascript`?
16:00:05	FromDiscord	<flywind> Nim JS backend doesn't support hash float.
16:00:41	*	narimiran joined #nim
16:02:27	Zevv	well, it's not that it does not support it. it's broken.
16:03:07	FromDiscord	<flywind> The version before returns floats, I don't think that's right
16:04:41	FromDiscord	<flywind> `import hashes; echo hash(0.3456)` returns `0.3456` in Nim v1.2.6 in JS backend.
16:10:33	FromDiscord	<flywind> I would fix cast float to int first.
16:19:25	Zevv	how does one reliably run all CI locally?
16:19:32	Zevv	./koch runCI?
16:24:44	FromDiscord	<flywind> run all tests?
16:25:02	Zevv	yeah seems I have nodejs versions issues
16:25:06	Zevv	so im good
16:26:39	FromDiscord	<flywind> https://nim-lang.org/docs/contributing.html#running-tests
16:27:12	Zevv	yeah, but that says "a good subset of"
16:27:33	FromDiscord	<Sorrow> sent a code paste, see https://play.nim-lang.org/#ix=2KvZ
16:27:35	Zevv	but I hate to fire github CI all the time, and I have a 28 core machine, so it makes sens to make sure i get it through clean before pushing
16:29:00	*	JustASlacker joined #nim
16:29:36	*	vivus joined #nim
16:30:18	FromDiscord	<Rika> sorrow, maybe use async
16:30:27	FromDiscord	<Rika> not sure if that would actually help though
16:30:40	FromDiscord	<Sorrow> I read about the progress callback thing.
16:41:42	*	NimBot joined #nim
16:47:47	*	pigmej joined #nim
16:48:05	pigmej	Hey guys,
16:48:17	pigmej	Anyone having experience using cligen/dents ?
16:48:39	pigmej	I'm in need of having recursed directory / file listing BUT I would like to filter some "expensive" directories before it starts to recurse there
16:49:08	pigmej	I know that there is preRec in cligen/dents BUT unless I'm mistaken I don't see a way to break / stop go to the directory
16:49:58	pigmej	I am aware of globs/walkDirRecFilter BUT there is a warning about `unstable API`
16:56:33	FromDiscord	<Clyybber> Zevv: Hmm, but actually we can just extend the lifetime of i ourselves, no?↵Since we are rewriting locals anyways
16:56:57	FromDiscord	<William_CTO> I'm using unittests and trying to run a test which has subtests, but the subtests aren't being ran with `nim c -r tests.nim "parent test"`
17:01:36	FromDiscord	<Rika> is it possible to make an object with a size of 0?
17:01:57	*	lritter joined #nim
17:02:19	FromDiscord	<lqdev> what for?
17:02:25	FromDiscord	<haxscramper> `distinct void`?
17:02:48	FromDiscord	<Rika> i want to store compile time data into another object without runtime cost (this is very fucky i know)
17:02:53	FromDiscord	<lqdev> !eval type O = object; echo O.sizeof
17:02:55	NimBot	1
17:02:57	FromDiscord	<Rika> 1
17:03:00	FromDiscord	<Rika> its gonna be 1
17:03:01	FromDiscord	<Rika> yeah
17:03:16	FromDiscord	<lqdev> eh mate it's only one byte
17:03:20	FromDiscord	<haxscramper> !eval type VOID = distinct void; echo VOID.sizeof
17:03:20	FromDiscord	<lqdev> it's nothing
17:03:23	NimBot	0
17:03:28	FromDiscord	<Rika> hm
17:03:33	FromDiscord	<Rika> lol nice
17:03:48	FromDiscord	<haxscramper> curios what it compiles to
17:04:28	Zevv	clyybber: the original var args might live in a non-cps proc that returns before the continuation is resumed
17:04:49	Zevv	it's perfectly possible I don't understand what disruptek is trying to say here, of course. I have that, sometimes
17:05:39	FromDiscord	<Clyybber> what you said makes sense, but im in the same boat as you, i dont understand the gh issue
17:05:54	Zevv	so, is it us, or is it disruptek, I wonder
17:08:09	FromDiscord	<Clyybber> i dont understand if its trying to tellme theres a bug I could fix
17:08:19	FromDiscord	<William_CTO> sent a code paste, see https://play.nim-lang.org/#ix=2Kwk
17:08:20	Zevv	I don't think so
17:08:30	Zevv	I think it's a CPS internal thing. Did he ping you on that one?
17:08:53	disruptek	what?
17:09:01	FromDiscord	<haxscramper> https://play.nim-lang.org/#ix=2Fbo
17:09:12	Zevv	does it hurt bad?
17:09:31	Zevv	clyybber was wondering about https://github.com/disruptek/cps/issues/47
17:09:31	disbot	➥ mutable params and the developers who love them
17:09:39	FromDiscord	<For Your Health> What would be a good way to organize the following: I have a trading bot that needs to interface with an exchange, and I have a simulated exchange that I'm writing. I want the bot to be able to use either the simulated exchange or the real exchange.
17:09:40	Zevv	I say it can't be done
17:10:01	FromDiscord	<For Your Health> I'm guessing I need dynamic dispatch and I haven't used that in Nim yet.
17:10:16	disruptek	it can be done.
17:10:32	Zevv	For Your Health: Emulate your exchange at the API level you are accessing it at in real life
17:10:49	Zevv	disruptek: who is calling foo(). A non-cps proc?
17:11:03	Zevv	that might have `c` on its stack. It might have returned before the continuations get resumed
17:11:07	Zevv	so `c` is dead & gone
17:11:09	Zevv	dust
17:11:10	disruptek	yes.
17:11:28	FromDiscord	<For Your Health> I'm more so meaning how do I interface with it from the bot's point of view? Does the bot hold a reference to an exchange object or something?
17:11:37	disruptek	probably.
17:12:00	disruptek	if it were me, i would mock the entire API.
17:12:09	disruptek	i presume you're using http, so just mock the server.
17:12:36	disruptek	then run it on a distant cloud, etc.
17:13:20	disruptek	zevv: and, so what?
17:14:13	FromDiscord	<William_CTO> How can I debug an iterator? I can't call repr on it and I'm getting `ObjectAssignmentDefect` when I try to iterate over it with `for k in`
17:14:31	disruptek	use debugEcho.
17:14:36	FromDiscord	<For Your Health> So should I have something like `Exchange = ref object of RootObj` and `SimulatedExchange = ref object of Exchange` ?
17:14:56	FromDiscord	<For Your Health> And then my bot holds a ref to Exchange and calls on that or something?
17:14:57	disruptek	no.
17:15:18	FromDiscord	<William_CTO> Error: attempting to call routine: 'keys'↵Same issue as trying to use repr and echo
17:15:27	disruptek	well, don't do that.
17:16:17	Zevv	disruptek: ok, nevermind, i must be stupid and you must be smart, so i give up
17:16:23	FromDiscord	<William_CTO> sent a code paste, see https://play.nim-lang.org/#ix=2Kwm
17:16:38	disruptek	zevv: i don't think i get the problem.
17:17:40	Zevv	what does c refer to if it was a local of the calling function, which returned before we *use* c
17:18:01	disruptek	probably a pointer. i don't know how else to impl it.
17:18:15	Zevv	right. and is this pointer still valid?
17:18:20	disruptek	nope.
17:18:24	Zevv	iright
17:18:31	Zevv	so why do we need to support this
17:18:40	disruptek	because people like mamy exist.
17:19:01	Zevv	if you want to do that kind of thing, use heap refs
17:19:34	Zevv	if mamy needs this, let him request it
17:19:57	Zevv	and show how this should actually work
17:20:08	disruptek	so the current blocker is kinda silly.
17:20:19	disruptek	we're not substituting all variables.
17:20:44	disruptek	the issue has something to do with scope. i haven't thought through where it has to be done. do you have an idea?
17:20:50	ForumUpdaterBot	New thread by NightRang3r: Executing command in hidden console, see https://forum.nim-lang.org/t/7320
17:21:03	disruptek	where it's done now doesn't work. 😉
17:21:40	disruptek	i think we're doing substitution on only a single statement instead of the remaining scope.
17:21:47	FromDiscord	<For Your Health> So should the simulated exchange and the real exchange be abstracted out to separate modules or something? I'm still confused on a good way to swap them out.
17:22:04	disruptek	the /right/ way to do it is to mock the entire server.
17:22:09	Prestige	Zevv: about my question about orc earlier, https://github.com/avahe-kellenberger/nimdow/blob/master/src/nimdowpkg/windowmanager.nim line 175. I get an error saying the cast is incorrect
17:22:14	Zevv	thats what I said
17:22:23	Prestige	But only when I use --gc:orc
17:23:13	disruptek	man, your use of whitespace is the stuff of nightmares.
17:23:39	disruptek	forget MI, i don't think you should be allowed access to the newline.
17:23:56	*	JustASlacker quit (Ping timeout: 240 seconds)
17:24:11	Prestige	Haha, I think it's much more readable to have params vertically aligned
17:24:36	FromDiscord	<Rika> y tho
17:24:48	FromDiscord	<For Your Health> I'm not sure what mock the entire server means. I'm working on the simulated exchange first and I am keeping it simple. So imagine you can ask the simulated exchange to put in a market order or whatever and it will simulate what that does to your capital and account for fees and stuff.
17:24:50	FromDiscord	<Rika> oh this style
17:25:03	FromDiscord	<Rika> i was thinking the one where the first param is not newlined
17:25:08	FromDiscord	<Rika> that kinda style is scarier
17:25:14	Prestige	Yeah for sure
17:25:25	FromDiscord	<Rika> true waste of space that one is
17:26:14	disruptek	health: how do you communicate with the real exchange?
17:26:35	Prestige	I also greatly dislike horizontal scrolling
17:26:53	FromDiscord	<For Your Health> I'm not communicating to it yet as far as putting real trades on it, I'm only reading prices off of it with WebSocket for backtesting
17:27:39	FromDiscord	<For Your Health> But where I get the prices from doesn't really make a difference at the moment. The simulated exchange is designed to just accept prices that you feed it
17:27:46	disruptek	so create your own websocket server. make it do the same shit the exchange does. to run your tests, you talk to this "mock" server instead of the real server, by simply altering the server address in your ws connection statement.
17:27:49	Zevv	ok, but what protocols, http?
17:28:00	Zevv	moch the same http api call in your fake server
17:28:15	Zevv	right, what he said ^^
17:29:30	FromDiscord	<For Your Health> The web protocols aren't really what I'm asking about at the moment. Getting the prices from the exchange, putting buy orders on the exchange, etc isn't what I'm having problems with. I want to know how I can structure my bot's code in such a way that it doesn't matter whether or not I'm using a real exchange or a fake exchange.
17:29:35	*	Arrrrrrrr joined #nim
17:29:45	Zevv	...
17:29:45	disruptek	and we just told you that.
17:30:15	Zevv	make it talk like a duck and walk like a duck
17:30:54	FromDiscord	<For Your Health> My question is more along the lines of how do you do that in Nim? Like in C++ or whatever I would do it with inheritance or something
17:31:05	FromDiscord	<For Your Health> So how do you duck type in nim I guess is what i'm asking?
17:31:05	disruptek	you're missing the point.
17:31:29	Zevv	yeah i see wha youre trying to do
17:31:41	Zevv	you want to mock at a different level
17:31:51	disruptek	the point is to write as little code as possible that varies between the two implementations.
17:32:16	Zevv	you can use generics or inheritence
17:32:19	disruptek	mocking the entire server achieves that because /only data/ changes, and /only one int32/ needs to be updated.
17:32:23	Zevv	generics is the 'nim way'
17:32:24	FromDiscord	<juan_carlos> If it talks like a duck and walks like a duck, is obviously a platypus.
17:34:15	FromDiscord	<For Your Health> So by mocking the entire server do you mean having a large module that handles both simulating an exchange and connecting to a real exchange, where you change some internal state to determine which one you are doing?
17:34:29	FromDiscord	<For Your Health> And my bot references that module?
17:34:33	disruptek	no.
17:35:01	FromDiscord	<For Your Health> I guess what I'm asking is, what does my bot call to place a buy order?
17:35:07	disruptek	i mean you go off and /write an exchange/ and then /run it somewhere else/ to mimic the actual reality.
17:35:17	disruptek	i give up.
17:35:44	disruptek	i have more stuff i want to get done in 2021.
17:35:55	FromDiscord	<For Your Health> I guess I'm not communicating my problem correctly, my apologies.
17:36:08	disruptek	you are. you aren't listening to the answer.
17:42:14	FromDiscord	<Rika> how would i go about making a macro that accepts a typedesc and gets the next smaller version of it? ex uint16 -> uint8
17:42:21	Prestige	Health: Basically you just need to change the URL of the api endpoint. You can pass the url to the construction of your program/object/whatever
17:43:12	Prestige	Everything else should stay the same, except for that
17:43:13	disruptek	rika: fairly manually, i'm afraid.
17:43:42	FromDiscord	<Rika> it's a pain to get this to work with a generic parameter
17:43:46	FromDiscord	<Rika> (i do not know how)
17:44:20	FromDiscord	<For Your Health> Alright I went back and reread the suggestions. I think I get what everyone is saying now. You are saying to have my simulated exchange be a server that I connect to with websocket that is essentially indistinguishable from the bot's point of view to a real exchange?
17:44:22	disruptek	you can measure the type easily enough using one of the three getType impls.
17:44:30	disruptek	health: yep.
17:44:54	FromDiscord	<Rika> imma refresh on getType then thanks
17:45:31	disruptek	rika: once you have the nnk enum, you can just resolve it from one of two arrays -- signed and unsigned. something like that.
17:46:02	*	JustASlacker joined #nim
17:47:13	FromDiscord	<Rika> i still have no idea how it'd do with generic params though... it still sees `T` and not whatever it should resolve to...
17:47:44	disruptek	of course.
17:48:01	disruptek	your macro won't use generics.
17:48:43	FromDiscord	<Rika> i mean when i feed my macro a generic param, what can i do D:
17:49:10	disruptek	do you have a repro?
17:50:21	FromDiscord	<Rika> one moment
17:50:27	disruptek	health: it sounds like a lot of work but it ultimately is a much higher-quality solution that sidesteps inumerable flaws with the other approaches.
17:50:49	*	hmmm quit (Quit: rebootan)
17:52:07	disruptek	it's work that will help you understand your client better, too.
17:55:20	FromDiscord	<Rika> disruptek: https://play.nim-lang.org/#ix=2KwB
17:56:44	FromDiscord	<For Your Health> Shouldn't I still have a layer of indirection somewhere? Like say I want to connect to two different exchanges, they won't both have the same API. My bot shouldn't care what exchange it's connected to for the most part.
17:57:01	FromDiscord	<For Your Health> Two different real exchanges that is
17:57:55	saem	Health: I start with when isMainModule. In your case I would start in a prototype bot module to just pass basic things to it get a sketch of the core.
17:58:17	saem	You don't want indirection.
17:59:02	saem	They're not different, it should be the same thing. The bot only differentiates based on address.
17:59:04	FromDiscord	<William_CTO> I figured out my issue disruptek
17:59:50	FromDiscord	<For Your Health> I have a decent core of the bot itself already. I was simulating the entire exchange from within the bot to develop it. I'm moving on the the point where I want to simulate an exchange externally that has delays when you request market orders.
18:00:14	FromDiscord	<William_CTO> I'm using json's `to` proc to unmarshal json into an object. Some of my objects were object of AnotherBase which was causing the issue.
18:01:43	FromDiscord	<William_CTO> @For Your Health what is this bot of yours?
18:01:45	Prestige	I think you'd want indirection if the market APIs are different, for that you could use inheritance or concepts (if concepts are working)
18:02:11	Prestige	But if the APIs are the same there's no point in doing that
18:02:28	saem	Then write the exchange, it should be behind an http client, not an http client and a mock http client. If you really want to mock it, then you could use generics or a compiler define and have a mock vs non-mock implementation. Our if you actually want them swappable at runtime then you need indirection.
18:03:04	saem	Otherwise you want One Direction
18:03:42	saem	Yeah, what Prestige said
18:03:58	FromDiscord	<For Your Health> It's a crypto trading bot I'm toying around with.
18:04:21	saem	Whatever.
18:04:32	saem	Does it need runtime swapping?
18:05:16	FromDiscord	<For Your Health> It doesn't need runtime swapping really, I just want to be able to backtest the bot on recorded data, which could be done with a different program
18:05:37	FromDiscord	<For Your Health> I want to bot's code to be generic in that it doesn't care whether or not it is connected to an exchange or backtesting
18:06:03	FromDiscord	<For Your Health> Like you feed it price data sequentially and it tells you when to buy and sell and how much to buy and sell
18:06:27	disruptek	then do it the way the old guys told you to do it.
18:06:28	disruptek	rika: i have no bandwidth, but did you get my comment?
18:06:43	*	JustASlacker quit (Ping timeout: 268 seconds)
18:08:46	saem	Then use a compile time mechanism to do that, run it as a separate thing, build your internal architecture as a log (if LMax is to be believed this is the one true way), use inheritance and methods like you would if you want run time. But I'm not sure what the question is anymore, because it sounds like you know.
18:09:12	FromDiscord	<ITR> I've noticed that there's multiple garbage collectors in nim, but is there any way to change how memory allocation is done?
18:09:30	mipri	change what about it?
18:10:05	FromDiscord	<ITR> IE. make objects of the same type be located near each other in memory
18:10:11	FromDiscord	<ITR> (edit) "memory" => "memory, no matter when allocated"
18:10:27	FromDiscord	<ITR> (edit) "allocated" => "created"
18:10:47	FromDiscord	<haxscramper> You probably need to write custom allocator for that one and dive into low-level C details
18:11:04	FromDiscord	<haxscramper> IIRC default GC does not provide such features
18:11:05	FromDiscord	<ITR> sent a code paste, see https://paste.rs/Acw
18:11:10	*	hmmm joined #nim
18:11:33	FromDiscord	<ITR> I mean, I feel like that's separate from GC, IE, a garbage _collector_ could work independently from an allocation manager
18:12:43	FromDiscord	<haxscramper> https://forum.nim-lang.org/t/3158
18:12:48	mipri	sure. --gc:arc has a -d:useMalloc option. something similar to that might get you a different allocator. but when people care about allocators like this they tend to want to pick the allocator for the object, not just pick an allocator for the entire program.
18:12:48	FromDiscord	<juan_carlos> GC wont alloc.
18:13:10	FromDiscord	<haxscramper> Maybe something from this tread would be useful, it seems like a question quite close to yours
18:13:47	saem	ITR of you use object and contiguous allocation structures like seq, does that not do what you want?
18:14:18	mipri	LeuGim's answer there is what I was thinking of as well, where use the allocator to get memory and then cast it to your object, but that requires explicit constructors
18:15:14	FromDiscord	<ITR> I'm looking into entity component systems (mostly gamedev related), use hashsets to store components for a specific object.
18:15:23	FromDiscord	<ITR> which use
18:15:28	FromDiscord	<Rika> disruptek, sorry, which comment do you refer to?
18:15:35	saem	I knew it.
18:15:39	disruptek	don't you want `auto` as your return-type?
18:15:48	FromDiscord	<Rika> Ah.
18:15:51	FromDiscord	<Jarred Kenny> Is it possible to get the position of a key in a table?
18:15:55	FromDiscord	<Rika> LOL
18:16:02	FromDiscord	<Rika> That is indeed a better idea
18:16:04	FromDiscord	<Rika> Thanks
18:16:08	disruptek	if you use an Ordered table, yes.
18:16:34	FromDiscord	<ITR> @Jarred Kenny what do you mean by position?
18:16:51	saem	ITR: look at the internals of a tables. It'll behave well enough for your case.
18:17:01	FromDiscord	<Jarred Kenny> Essentially the index from the order of insertion
18:17:09	FromDiscord	<haxscramper> IIRC jken was doing the same (also ECS, also need for continuous memory alignment)
18:17:21	FromDiscord	<Jarred Kenny> lol yes, still playing with this.
18:17:37	FromDiscord	<ITR> saem: even if they point to an object/ref type?
18:17:59	FromDiscord	<ITR> though I guess they don't technically have to be ref types in the table
18:18:17	saem	Why are you using a ref? That's basically exactly not what you want for a component
18:18:28	FromDiscord	<ITR> yeah, I just realized that
18:18:54	FromDiscord	<haxscramper> As suggested above you might want to look into implementation details for `OrderedTable`, but there is no default API to get this AFAIK
18:19:41	saem	You're _doing_ the larger scale memory layout, use object.
18:20:16	FromDiscord	<ITR> I was looking at cat-400 which was talking about ref object, and I felt like it did ecs wrong, so I looked at some opinions people had about other ecs implementations, but most other languages have ref type by default for objects which made me forget, lol
18:20:21	saem	Also, get used to not using any refs and indirecting based on IDs
18:20:31	FromDiscord	<ITR> It won't work for what he wants, I think
18:21:03	saem	Yeah, checkout the three other ECS implementations on nimble
18:21:24	FromDiscord	<ITR> @Jarred Kenny If you insert 5 objects, and remove the middle one, do you then want the "index" of each object to be 1, 2, 3, 4 in order of execution?
18:21:50	FromDiscord	<Jarred Kenny> The more I think about it, the more I feel there is a better approach.
18:22:10	FromDiscord	<ITR> order of insertion
18:22:15	FromDiscord	<ITR> @Jarred Kenny what do you want to use it for?
18:23:32	FromDiscord	<ITR> You _could_ have each contain an "id" which starts at 0 and always increases, then have a separate binary tree which keeps track of which ids are still in use, then get the index there, but that sounds very complex and probably isn't needed
18:23:43	saem	Also if your memory requirements are big and your specific component density is high, use a plain seq for entity and a same sized seq for the components. Then you ID is the same and it's even simpler.
18:25:30	FromDiscord	<ITR> hmm, I only see two others, akrsys and entoody
18:25:50	saem	You don't need to do that either, unless you have very specific indexing needs. Simply have a sequence for you component object. The sequence index is the component identifier. Then have a table which maps entity identifier to component identifier.
18:27:35	saem	!repo ecs
18:27:36	disbot	https://github.com/yglukhov/ecs -- 9ecs: 11Entity Component System 15 21⭐ 2🍴 7& 7 more...
18:27:52	FromDiscord	<ITR> o nice
18:27:53	disruptek	!repos ecs
18:27:55	disbot	query failed 😢
18:27:57	disruptek	!repos ecs
18:27:59	disbot	query failed 😢
18:28:05	FromDiscord	<ITR> !repo, not repos
18:28:07	disruptek	wtf github
18:28:12	disruptek	!repos ecs
18:28:13	disbot	query failed 😢
18:28:35	disruptek	i think they put new query limits in place.
18:28:58	saem	Butts
18:29:35	saem	I forgots the 's', now we can't have nice things. :(
18:29:53	disruptek	anyway, yuri's is a toy. you want polymorph or pixecs.
18:30:00	saem	Yeah
18:30:25	disruptek	`repo` delivers the top hit. `repos` delivers the top 3, or the top 30 if you /msg disbot directly.
18:30:52	FromDiscord	<ITR> neither polymorph nor pixecs returns any results on nimble.directory
18:31:10	saem	You and the dms, you sly dog
18:31:28	FromDiscord	<ITR> found on google tho
18:31:36	saem	The system works
18:31:38	disruptek	nimble.directory is meaningless.
18:31:39	Zevv	is disbot hotpluggable?
18:32:03	disruptek	no.
18:32:07	Zevv	lame.
18:32:38	saem	Ssshhh, future robot overlords can hear you.
18:32:44	FromDiscord	<haxscramper> There are almost 4k nim repos on github
18:33:50	saem	Yup, I have a search I revisit to see latest by activity. Shows some interesting of work. AoC basically DDoSed that for a bit.
18:33:53	FromDiscord	<ITR> why do all the ecs repos start with an argument for why you might want to use them
18:34:04	saem	Because it's ECS
18:34:09	saem	It's like crossfit
18:34:10	mipri	why do any repos fail to start with such a thing?
18:34:20	saem	You're not doing it unless others know you are
18:34:43	saem	Also, I'm paleo
18:35:12	saem	Seriously though, the why is good. Wish more things did that.
18:36:22	FromDiscord	<ITR> i'm very tempted to make a frankenstein like project with the cat-400 framework, but using rapid's opengl wrapper aglet, and polymorph for ecs
18:36:58	*	vivus quit (Ping timeout: 272 seconds)
18:36:59	saem	One of my favorite Nim repos is synthesis by mratsim. Great explanation, nice layout, code has lots of explanations.
18:37:19	saem	ITR that sounds fun
18:38:59	FromDiscord	<ITR> :​o synthesis has state machines?
18:39:04	Zevv	19:38 < FromDiscord> <ITR> :​o synthesis has state machines?
18:39:11	Zevv	synthesis *is* state machines
18:39:28	FromDiscord	<ITR> that's literally the last thing i need, adding it to frankenstein
18:39:29	disruptek	you're blowing my mind right now.
18:39:43	Zevv	always will, mate
18:39:54	disruptek	why doesn't tzevv work?
18:40:30	reversem3	Does anyone know of server modules for video streaming like nginx RTMP modules?
18:40:35	Zevv	disruptek: gimme a sec
18:41:02	Zevv	pull
18:41:21	Zevv	oh dang no
18:41:26	Zevv	sheit
18:41:38	disruptek	it compiles, at least.
18:41:40	FromDiscord	<ITR> strange, Polymorph isn't on nimble, but it has a .nimble file
18:41:45	Zevv	disruptek: yeah
18:42:09	Zevv	I got the first 2 or 3 up
18:42:17	Zevv	and the rest all fails on expPrims asserts
18:42:44	FromDiscord	<ITR> is there any way I can pass the github link to nimble?
18:42:45	saem	Nimble requires some hoops (not bonkers) to register your stuff and not everyone does it. I mean I still can't be arsed.
18:43:38	disruptek	zevv: did you have to fix cps?
18:44:02	saem	Yes, it's either install command with a flag or development. Check nimble --help
18:44:02	disruptek	i don't publish on nimble because the publish process is stupid and dom wouldn't accept my patch to fix it.
18:44:21	Zevv	disruptek: no i only changed the .cps. syntax and switched to testes
18:44:37	disruptek	i guess that's good.
18:44:41	Zevv	but the tests themselves fail so i think something is bonkers in the xfrm
18:45:11	Zevv	i planned to do some stuff today but i ended up doing set[char].repr in the nim libs
18:45:14	disruptek	"no generic params allowed for ref" from environment.nim.
18:45:28	Zevv	what does that even mean
18:46:00	Zevv	im on mobile now, cant take a loook sorry
18:46:32	disruptek	the bad line is from populateType, so it's due to assuming y is a constant in `var x = y`.
18:47:06	Zevv	9
18:48:03	disruptek	we don't use the `cps` prefix, do we?
18:49:43	Zevv	like i said, mentally im still stuck at 0.0.13
18:49:43	disruptek	result = env_33554996(fn: t, result_33554995: result_33554995)
18:49:46	disruptek	glwt
18:50:56	disruptek	any problems with my testes? other than the fact that they don't work for you, i mean?
18:51:19	Zevv	oh i just need to figure out how it works i guess
18:51:43	disruptek	the transform is pretty weird. i dunno if it's your fault.
18:51:51	Zevv	'compilation failed' is cool and all, but i think i would like it to just barf and stop at the first occurnce so i can fix it
18:51:54	disruptek	60 proc C(continuation: env_33554996): C {.cpsLift.} =
18:51:54	disruptek	61 env_33554996(continuation).result_33554995
18:52:23	disruptek	you just run it with --define:release to make it barf.
18:52:36	Zevv	ah right you told me
18:53:06	disruptek	but maybe i can make that work somehow, too.
18:53:51	disruptek	i didn't expect that C() proc, either.
18:53:53	*	vivus joined #nim
18:55:54	Oddmonger	that must be the magic of the new year, but the manual seems clearer now :)
18:58:27	saem	Oddmonger: I've had a bunch of those moments myself recently. :D
18:59:02	ForumUpdaterBot	New thread by Neodim: Native GUI development for MacOS, see https://forum.nim-lang.org/t/7321
19:00:20	FromDiscord	<Daniel> Happy New Year 🥳
19:00:25	FromDiscord	<Daniel> https://media.discordapp.net/attachments/371759389889003532/794641193287942144/2020.gif
19:00:52	disruptek	zevv: i dunno, tzevv code looks okay to me.
19:01:05	Zevv	i thought so as well
19:01:22	disruptek	and those transforms look pretty bogus.
19:02:00	disruptek	oh, i see it.
19:02:12	disruptek	t(): C {.cps: C.} should be t() {.cps: C.}
19:02:44	disruptek	i pushed.
19:02:45	Zevv	pfff
19:03:10	Zevv	silly, we should hide that stuff and just expose a few friendly macros instead
19:03:34	disruptek	we do hide it.
19:03:48	Zevv	so, now the first few run and the rest assert, right?
19:03:50	*	junland quit (Quit: %ZNC Disconnected%)
19:03:51	disruptek	your t() doesn't have a return type.
19:03:55	Zevv	because that's how far I got before
19:04:10	disruptek	## 21 tests 🟢8 🔴5 ⛔8
19:04:17	*	superbia joined #nim
19:04:45	disruptek	we have a duplicate proc predeclaration, which is funny.
19:04:53	*	junland joined #nim
19:05:01	disruptek	i love having these bugs, though.
19:05:09	disruptek	so easy to fix one way or another.
19:05:18	disruptek	i really hate debugging the compiler.
19:05:33	Zevv	this is not the compiler being wrong
19:05:35	Zevv	its just cps
19:05:44	disruptek	i know, that's what i like about it.
19:05:48	Zevv	ah
19:06:12	disruptek	like, i know that if i put a couple hours in, i will pass more tests.
19:06:15	disruptek	that's a nice feeling.
19:06:22	Zevv	go for it
19:07:33	*	blueberrypie quit (Quit: leaving)
19:08:16	*	junland quit (Client Quit)
19:09:57	*	hnOsmium0001 joined #nim
19:10:08	*	junland joined #nim
19:12:53	*	superbia quit (Quit: WeeChat 3.0)
19:20:54	Zevv	so we can start dropping the :int parts soon, right?
19:23:49	*	himu joined #nim
19:23:58	disruptek	we already do.
19:24:12	Zevv	well look at that
19:24:22	FromDiscord	<Avatarfighter> Hello hello
19:24:26	FromDiscord	<Avatarfighter> what is CPS being used for?
19:24:32	Zevv	cmon dude
19:24:51	disruptek	i updated the README with a new typed tock.
19:25:07	Zevv	avatarfighter: nothing *yet**
19:25:37	Zevv	but the idea is to make it the lego bricks for building anything involving non-linear code flow
19:26:14	Zevv	like custom control flow, coroutines, async, lazy eval, efficient fp
19:26:49	leorize	I can build you a new async system or port the current one over when y'all are done with the cps stuff
19:27:10	disruptek	you may as well start now.
19:27:50	leorize	I heard there are still a bunch of bugs :P
19:27:56	disruptek	zevv: what do you make of my hiddenstdconv hack?
19:28:11	leorize	and I'm writing that process library again so will take awhile
19:28:14	Zevv	dont know what that is
19:28:51	disruptek	https://github.com/disruptek/cps/commit/ee4c7a8240c02c2c281fe67fb2e339b6f17a218d
19:29:47	FromDiscord	<Avatarfighter> leorize: new async piques my interest 😛
19:29:53	*	vivus quit (Quit: Leaving)
19:30:33	Zevv	hiddenStdConv, new to me
19:30:49	*	vivus joined #nim
19:32:07	disruptek	i was thinking maybe we should do the transform at a lower level.
19:34:38	*	wekan joined #nim
19:34:46	*	wekan quit (Remote host closed the connection)
19:36:33	*	xet7 joined #nim
19:38:15	*	vivus quit (Quit: Leaving)
19:39:40	Zevv	hmm like lower how?
19:41:21	*	a_chou joined #nim
19:41:49	disruptek	like making getImplTransformed work.
19:44:53	disruptek	you feel me?
19:46:11	*	a_chou quit (Client Quit)
19:49:15	Zevv	that would be sweet indeed
19:49:38	Zevv	I was just wondering, is it solid enough?
19:49:43	Zevv	the docs say it's for inspection and debugging
19:49:49	Zevv	so are we supposed to actually /use/ that?
19:50:31	Zevv	Ah, this one, I remember now: "Error: implementation of 'exit_385877917' expected"
19:50:45	disruptek	just the duplicate predec.
19:50:52	Zevv	ah k
19:52:10	disruptek	getImplTransformed would let us support for loops.
19:52:18	Zevv	yes, that's why I looked into it
19:52:27	Zevv	but it works inside out
19:52:30	Zevv	it's funny
19:53:36	disruptek	inside out?
19:53:48	Zevv	you pass it a sym and it gives the implementation
19:53:55	Zevv	it's not like a macro that actually transforms your code
19:53:56	disruptek	yeah.
19:54:14	disruptek	but that's okay. we can maybe layer cps.
19:56:06	Zevv	I'm dropping testes. Everything is a green square.
19:56:23	disruptek	dude, my testes dropped along time ago.
19:56:32	disruptek	very long.
19:57:07	Zevv	can't see what its doing
19:57:46	disruptek	that's funny, because i wrote testes because i couldn't see what tzevv was doing.
19:58:10	*	himu quit (Quit: Konversation terminated!)
20:00:01	*	himu joined #nim
20:01:07	FromDiscord	<Deleted User 5bd78114> Imagine having two other people in your head who constantly give you ideas of what to code and point out common errors or logic errors that you've missed-
20:01:10	ForumUpdaterBot	New thread by Alexeypetrushin: Weird how visibility scope works if generic proc used. Possible bug?, see https://forum.nim-lang.org/t/7322
20:05:10	ForumUpdaterBot	New thread by Domogled: How use custom event inherited from DOM Event in javascript?, see https://forum.nim-lang.org/t/7323
20:06:43	disruptek	zevv: getImplTransformed /almost/ works.
20:07:50	saem	Hah, ran into a case where CPS would be handy (in JS land).
20:11:09	*	Arrrrrrrr quit (Quit: Arrrrrrrr)
20:13:15	saem	It's similar to a threadlocal perhaps. I have a bunch of code that calls console.log for logging, of course. Assuming I wrap console.log, I can have it look up an "actionId" so I can correlate what logical user/program action something belongs to. In order to do that I need a threadlocal-ish thing and I believe a dispatcher could manage setting/unsetting it as necessary.
20:13:49	*	hmmm quit (Quit: WeeChat 3.0)
20:13:59	saem	This is also a fun test to see if I understand what/how dispatchers could be used.
20:14:43	disruptek	i don't get it.
20:15:22	disruptek	if you wrap the log call, why wouldn't you pass the source of the call right there?
20:17:29	saem	It's fair question,t he wrap was going to be nothing more than a static template to lookup a known var.
20:18:04	Zevv	disruptek: already
20:18:25	disruptek	the transform is broken, afaict.
20:18:49	disruptek	the compiler gives me `var i`; no type, no initialization value.
20:19:01	disruptek	other than that, i think it would work.
20:19:49	Zevv	is there a way to tell what matches if you look from the inside and from the getImplTransformed side?
20:19:55	Zevv	can the nimnodes be ==ed ?
20:20:08	Zevv	if we are brave, we might be able to merge both views
20:20:12	Zevv	but that'll be tricky and fragile
20:20:15	disruptek	lemme push this. yeah, we could rewrite the rewrite.
20:20:32	disruptek	it's in a branch called `almost`.
20:20:57	Zevv	well that was pretty trivial code
20:21:09	disruptek	i would rather have the compiler fix its output, though.
20:21:20	Zevv	yeah sure, this is way more usable
20:21:23	saem	Even if I do wrap it, I'd have to have change each proc to accept a logger and pass it along (this could be rather long chains) rather than drop in a console.log call (or something incredibly similar) where I need it.
20:21:36	Zevv	i always felt we needed to go into the compiler to be able to get for loops expanded
20:21:45	Zevv	but if this works, that would be pretty nice
20:21:52	Zevv	so, yes, clyybber, we have work to do :)
20:22:07	disruptek	splash some cold water on his face.
20:22:25	Prestige	What do you mean by expanding for loops?
20:22:39	Zevv	CPS rewrites nim code
20:22:49	Zevv	but we can't transform non-closure for loops, becuause we can't see "inside"
20:23:08	Prestige	ah
20:23:09	Zevv	but there'a hacky macros proc getImplTransformed() that allows you to view a proc post-transform
20:26:20	FromDiscord	<Jarred Kenny> @haxscramper If you are curious, this is what I've ended up with for storing components in my ECS https://play.nim-lang.org/#ix=2KxI
20:27:34	FromDiscord	<Jarred Kenny> There is some inheritence at play with AbstractComponentList, but the components are stored in contigious seqs. Some sorting will be required,
20:28:22	FromDiscord	<Jarred Kenny> important part is met, which is components are not ref types, i think!
20:28:50	*	himu quit (Quit: Konversation terminated!)
20:29:28	*	himu joined #nim
20:29:52	*	himu quit (Client Quit)
20:30:10	*	himu joined #nim
20:30:53	*	himu quit (Client Quit)
20:40:12	saem	Jarred any reason to have a list of all the lists of components and not an object with one field for each component sequence?
20:42:51	FromDiscord	<Jarred Kenny> seam, the idea behind is you can use any object as a component and register them dynamically
20:43:11	FromDiscord	<Jarred Kenny> rather than having all possible component types defined in the Universe object
20:44:24	*	tane quit (Quit: Leaving)
20:49:09	saem	Hmm, you mean dynamic in so far as a module somewhere else, but still in the project can add in component types or possibly across HCR?
20:50:16	jken	not sure what HCR is, the idea is that someone consuming my code as a module could use arbitrary objects as components in the ECS
20:51:07	saem	jken: hot code reloading. If they're consuming your module then a generic type would work, wouldn't it?
20:51:58	jken	How would that work?
20:57:10	saem	If you have a Universe[C], where C is Component, then for add/remove it's concepts or bunch of templates. Probably can also do it via inheritance of your abstractcomponentlist and implement some methods.
20:57:26	saem	C is Components (the list)
21:11:58	*	narimiran quit (Ping timeout: 256 seconds)
21:12:42	FromDiscord	<William_CTO> I'm looking for a way to create a macro to generate `Option[x] = none(x)` for use in procs eg: `proc foo(bar: myMacro(string)`. I've never used macros before. Can someone point me in the right direction?
21:12:49	FromDiscord	<William_CTO> I've tried this so far (doesn't work)
21:12:55	FromDiscord	<William_CTO> sent a long message, see http://ix.io/2KxS
21:13:08	FromDiscord	<William_CTO> (edit) "long message," => "code paste," | "http://ix.io/2KxS" => "https://play.nim-lang.org/#ix=2KxT"
21:14:16	*	narimiran joined #nim
21:16:33	*	opal quit (Remote host closed the connection)
21:17:05	*	opal joined #nim
21:19:15	Zevv	disruptek: could it be that the underlying AST is ok but repr is not up to the task of properly displaying it?
21:19:30	disruptek	no.
21:20:08	jken	saem, concepts are still foreign to me.
21:20:16	jken	Can I bother you for an example?
21:20:34	Zevv	disruptek: how can you tell
21:20:44	disruptek	because my code sees all.
21:20:46	Zevv	because "var i" is just not legal AST, probably
21:21:09	Zevv	what does that look like in a treerepr?
21:21:18	disruptek	var (i, nil nil)
21:21:25	Zevv	:(
21:21:27	disruptek	yes.
21:21:44	FromDiscord	<William_CTO> dis can you suggest an approach if its even possible?
21:21:57	Zevv	yeah, well, that was my point: the docs say it's for inspection and debugging, not for real use for real things by grown ups
21:22:20	saem	jken: you've already read this: https://nim-lang.org/docs/manual_experimental.html#concepts
21:22:27	disruptek	william: what?
21:22:52	disruptek	i always compose ast "manually".
21:24:30	jken	saem, yeah I mean as they relate to the ecs
21:27:41	disruptek	william: you can, like, dumpAst or something to see how to compose it yourself. i just read the macros doc to see the structure and then recreate it.
21:28:08	saem	jken: https://github.com/disruptek/jason/blob/master/jason.nim#L4 shows anything that conforms to the predicate jason is and also covers arrays, so long as you can iterate over them and their elements are jasonable. You'd do the same for your component interface testing for add/remove procs
21:33:40	disruptek	gah who wrote that dogshit.
21:33:55	disruptek	oh wait.
21:43:47	Zevv	man git-dogshit
21:44:11	disruptek	No manual entry for git-dogshit
21:44:45	Zevv	oh, wait, "blame" it's called, I remember now
21:45:43	*	narimiran quit (Ping timeout: 246 seconds)
21:45:48	Zevv	Zzz. some caching up to do.
21:46:00	disruptek	aight, gn zevv
21:50:38	FromDiscord	<Clyybber> damn, just when zevv goes
21:50:54	FromDiscord	<Clyybber> disruptek this is what https://github.com/disruptek/cps/issues/48 is about?
21:51:04	disbot	➥ for loops could be supported with a compiler fix ; snippet at 12https://play.nim-lang.org/#ix=2Ky0
21:51:20	disruptek	i dunno what `this` refers to.
21:51:29	disruptek	oh, yes, is the answer.
21:51:34	FromDiscord	<Clyybber> he
21:51:53	FromDiscord	<Clyybber> does i have its .typ set?
21:51:58	disruptek	no.
21:52:06	FromDiscord	<Clyybber> Huh
21:52:32	disruptek	well, it's ast w/o a :type and i'm assuming if it has typ, it's an easy fix.
21:52:53	disruptek	i haven't tried to rewrite it myself in the macro.
21:52:59	FromDiscord	<Clyybber> yeah, should be relatively easy (probably)
21:53:08	disruptek	i agree.
21:53:20	disruptek	we could just lift it, but i think the compiler should do it for us.
21:53:25	FromDiscord	<Clyybber> I'll get to it in a bit
21:54:20	disruptek	cool, thanks.
22:02:16	disruptek	looks like ic is about to be merged.
22:09:11	FromDiscord	<Clyybber> neat
22:10:19	*	Vladar quit (Quit: Leaving)
22:13:18	*	xet7 quit (Remote host closed the connection)
22:23:10	disruptek	nimph rebuilds in 11s on ic and 17s w/o ic.
22:23:45	disruptek	oh wait, 10s w/o ic. hmm.
22:25:10	disruptek	yeah, no change in speed i guess.
22:25:13	*	PMunch joined #nim
22:25:16	*	natrys quit (Ping timeout: 246 seconds)
22:27:59	disruptek	testes is 0.10s to 0.15s slower under ic. changing a line and recompiling is 2x slower with ic than without.
22:28:41	FromDiscord	<Avatarfighter> It would be nice to see how IC affects much bigger projects
22:28:59	disruptek	much bigger than 550,000 lines?
22:29:16	FromDiscord	<Avatarfighter> well, I meant more other big projects
22:29:30	FromDiscord	<Avatarfighter> just to see if there is a change in compilation speeds
22:29:35	FromDiscord	<Avatarfighter> (edit) "just to see if there is a ... change" added "net"
22:30:03	FromDiscord	<Avatarfighter> Though its a bit concerning if there isn't an increase in compilation speeds on your projects disruptek, they are quite large usually
22:30:22	disruptek	testes is 70k lines, nimph is 553k.
22:30:27	FromDiscord	<Avatarfighter> yeah
22:31:05	disruptek	testes is ~0.55s w/o ic, with gc:arc and danger. 0.65-70s with ic, gc:arc, danger.
22:31:30	FromDiscord	<Avatarfighter> that's the bit that is concerning to me lol your projects are already large and if IC doesn't seem to have an effect on them then I wonder if IC will benefit projects of similar size
22:33:07	*	Gustavo6046 quit (Ping timeout: 268 seconds)
22:33:35	disruptek	nimp's c output is 46mb, the object files total 19mb, the rod (ic) totals 218mb.
22:33:41	disruptek	nimph, too.
22:37:16	disruptek	okay, i got it to go faster.
22:37:51	disruptek	i did a compile with the old compiler to blow away the cache -- 38s. and then again with ic to re-use the rod files -- 13s.
22:38:06	*	natrys joined #nim
22:38:55	disruptek	ic uses ~10% more memory, but it's clearly faster for some scenarios. the memory usage will improve, as will the speed.
22:41:04	FromDiscord	<Avatarfighter> Oh that's good I kind of feared IC was slower overall than before ngl
22:41:05	disruptek	compression would help. 218mb->17mb with gz.
22:41:49	*	natrys quit (Client Quit)
22:43:07	disruptek	the compiler builds faster without it, too. so far, i'm not impressed.
23:06:51	PMunch	How's everyone doing so far in 2021?
23:08:16	FromDiscord	<ElegantBeef> It's a new year so there has to be a new rampant disease to take over right?
23:10:01	PMunch	Well I mean we have the new strains going around
23:10:04	PMunch	Those seem fun
23:10:21	PMunch	But disease is so passé, we need something new for 2021
23:10:55	FromDiscord	<fwsgonzo> is it possible to call a varargs C function?
23:11:03	FromDiscord	<ElegantBeef> `{.varargs.}`
23:11:33	FromDiscord	<ElegantBeef> https://nim-lang.org/docs/manual.html#foreign-function-interface-varargs-pragma
23:12:04	*	Juno[m] joined #nim
23:12:47	PMunch	By the way, anyone else than me who have submitted a talk folk the virtual FOSDEM this year?
23:13:50	FromDiscord	<ElegantBeef> Jeez you butchered that sentence, and that's coming from me
23:14:02	FromDiscord	<fwsgonzo> thanks
23:14:13	PMunch	Give me a break, I'm recovering from Christmas/New years :P
23:14:22	FromDiscord	<lqdev> disruptek: is IC ready for some initial testing? i could try it on my projects
23:14:31	PMunch	And I haven't really spoken/written english for weeks at this poit
23:14:40	FromDiscord	<fwsgonzo> I think I give up - the best way to continue is to send and receive a serialized blob instead
23:15:21	PMunch	@fwsgonzo, what are you trying to do?
23:15:23	disruptek	lqdev: yes, try the araq/araq-ic5 branch.
23:15:35	disruptek	to use it, pass --incremental:on.
23:15:37	FromDiscord	<fwsgonzo> what I really want is to take a function as an argument, then enforce that the following arguments match the arguments that would go to this function, and then forward the arguments (after function) to a C function
23:16:11	PMunch	Hmm, not quite sure I understand
23:16:28	FromDiscord	<lqdev> disruptek: on latest devel?
23:16:40	disruptek	no, the araq/araq-ic5 branch.
23:16:43	FromDiscord	<ElegantBeef> Disruptek told you which branch
23:16:46	FromDiscord	<lqdev> kay
23:16:50	FromDiscord	<lqdev> will try tommorow
23:17:05	disruptek	and don't blame me for the performance. araq threw out most of my code.
23:17:06	FromDiscord	<lqdev> @ElegantBeef ah didn't see it
23:17:08	FromDiscord	<fwsgonzo> does the frozen serialization thing work on any object?
23:17:13	disruptek	yes.
23:17:24	FromDiscord	<fwsgonzo> can I serialize to stack?
23:17:27	disruptek	yes.
23:17:39	FromDiscord	<fwsgonzo> alright, thatll do it
23:18:00	disruptek	we don't yet raise if you try to serialize something stupid.
23:18:31	disruptek	i will probably add that for v1.0 when i reorder the api arguments for better `with` support.
23:18:33	FromDiscord	<dom96> ouch, if IC turns out to be slower than non-IC then that will be quite a let down
23:19:16	disruptek	i'm so glad i don't care.
23:19:40	disruptek	it would be really sad to put a ton of work into something that actually mattered and then have it be so marginal.
23:23:00	FromDiscord	<fwsgonzo> nimble install frosty?
23:23:27	disruptek	maybe?
23:23:34	disruptek	do you have access to the README.md?
23:23:58	FromDiscord	<fwsgonzo> yes checked the .nimble too, and `nimble search frosty` shows nothing
23:24:09	disruptek	what does the README say about Installation?
23:24:24	FromDiscord	<fwsgonzo> right, a github repo link
23:24:30	disruptek	ah, neat.
23:24:32	FromDiscord	<ElegantBeef> Disruptek doesnt use the nimble repository since he's too cool for it
23:24:45	disruptek	nimble is for fools.
23:24:48	FromDiscord	<ElegantBeef> Like so cool he made frosty
23:24:52	FromDiscord	<ElegantBeef> The registry isnt
23:26:22	disruptek	scramper just pointed out that github has 3500 more nim repos than nimble.
23:26:34	FromDiscord	<fwsgonzo> how common is import os?
23:26:47	FromDiscord	<ElegantBeef> well it should be `std/os` but fairly common 😄
23:27:09	FromDiscord	<fwsgonzo> guess im pretending its linux 🙂
23:27:10	FromDiscord	<ElegantBeef> I dont see how that is a good argument for not adding to a central repository
23:27:24	disruptek	i already explained my reasons once this year.
23:28:17	FromDiscord	<ElegantBeef> What was your suggested fix?
23:28:18	disruptek	if it helps you, feel free to add frosty to any registry you want.
23:28:49	disruptek	https://github.com/nim-lang/nimble/pull/728
23:28:51	disbot	➥ useful git commit messages during publish ; snippet at 12https://play.nim-lang.org/#ix=2fCa
23:32:13	FromDiscord	<ElegantBeef> This is sincerely your only issue with the registry?
23:32:57	disruptek	no, it's a stupid idea for other reasons.
23:33:44	disruptek	i don't tend to support software that sucks. if that makes me a bad person, well, i don't think this surprises anyone here.
23:33:49	FromDiscord	<ElegantBeef> Having easy access to published packages for quick development without having to search N number of git hosting services, seems like a pretty compelling case to me
23:33:59	disruptek	by N do you mean, 1?
23:34:12	disruptek	this is you reason for supporting shitty software?
23:34:25	FromDiscord	<ElegantBeef> Github, gitlab, src.ht, i've seen repos on nimble link to all of them
23:34:30	disruptek	it's literally garbage.
23:34:33	FromDiscord	<ElegantBeef> It doesnt matter your view on Nimble the software
23:34:35	disruptek	URLs work fine.
23:34:42	FromDiscord	<ElegantBeef> What's wrong with the god damn repository
23:34:47	disruptek	what's wrong with you?
23:34:52	disruptek	why do you care so much?
23:35:11	FromDiscord	<ElegantBeef> Cause i'm interested in why you avoid the repository with the excuse "software bad"
23:35:31	disruptek	i already explained it.
23:36:04	disruptek	if you want a deeper explanation, i wrote 8,000 lines of logic to explain how i feel open packaging should work in nim:
23:36:07	disruptek	!repo nimph
23:36:08	disbot	https://github.com/disruptek/nimph -- 9nimph: 11Nim package hierarchy manager from the future 🧚 15 91⭐ 5🍴 7& 1 more...
23:36:24	disruptek	if that's not enough for you...
23:36:29	disruptek	shit man, i don't know what to tell you.
23:36:35	FromDiscord	<ElegantBeef> Again i dont care about the package manager
23:36:41	disruptek	cool, bro.
23:36:45	FromDiscord	<ElegantBeef> A central repository is all we're talking about
23:36:55	disruptek	yeah, it's dumb.
23:37:12	FromDiscord	<ElegantBeef> You make such a compelling argument i've converted
23:37:16	disruptek	good.
23:38:15	FromDiscord	<haxscramper> I think someone just needs to add these ten lines of json diff to registry
23:38:31	FromDiscord	<haxscramper> Just for people to start annoying disruptek
23:38:50	disruptek	that started long ago.
23:39:10	disruptek	i was born annoyed and it just went downhill from there.
23:41:34	disruptek	my dist repo is the same idea implemented in a better way.
23:41:38	disruptek	!repo disruptek/dist
23:41:39	disbot	https://github.com/disruptek/dist -- 9dist: 11a nim distribution 👑 15 6⭐ 0🍴
23:43:24	planetis[m]	if you need a unique string/id each time an app is run, what would you use guys?
23:43:37	FromDiscord	<fwsgonzo> UUID perhaps?
23:43:43	planetis[m]	oids?
23:43:44	disruptek	std/monotimes
23:52:15	planetis[m]	thanks