<< 28-07-2022 >>
00:01:12FromDiscord<!Patitotective> seems like mariaDB and postgreSQL are also an option
00:01:32FromDiscord<dom96> because sqlite lives in a file, therefore it's much easier to deploy and maintain
00:02:12FromDiscord<dom96> you don't need mysql/postgre/etc unless you are a public tech company
00:02:26FromDiscord<dom96> even startups can do well with just sqlite
00:02:55FromDiscord<!Patitotective> its because the its an aws server↵but ill take your advice into account
00:03:07FromDiscord<!Patitotective> (edit) removed "the"
00:06:37FromDiscord<dom96> I'd also advise against aws lol
00:09:51FromDiscord<Sun「無用」> How do people make some functions that return a value... but has no return in the func? Like, having nothing at the end of the function
00:10:32FromDiscord<Sun「無用」> Or an implicit return, rust-like
00:11:02FromDiscord<dom96> !eval proc foo(): string = "hello world"; echo(foo())
00:11:05NimBotCompile failed: /usercode/in.nim(1, 22) Error: expression '"hello world"' is of type 'string' and has to be used (or discarded)
00:11:05FromDiscord<Elegantbeef> Nim has implicit results
00:11:19FromDiscord<dom96> oof
00:11:33FromDiscord<Elegantbeef> Point being expressions return the value
00:11:34FromDiscord<Sun「無用」> In reply to @dom96 "!eval proc foo(): string": Not that, ik that returns
00:11:44FromDiscord<Elegantbeef> or assignment of `result`
00:11:53FromDiscord<Sun「無用」> what's result...
00:12:08FromDiscord<Elegantbeef> Result is an implicit result variable
00:12:15FromDiscord<dom96> In reply to @NimBot "Compile failed: /usercode/in.nim(1, 22)": oh, that's just because it's single line, honestly not sure why it fails
00:12:43FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45RV
00:12:58FromDiscord<Sun「無用」> what...
00:13:18FromDiscord<Elegantbeef> !eval (proc foo()\: string = "hello world"); echo(foo())
00:13:21NimBotCompile failed: /usercode/in.nim(1, 10) Error: expected: ')', but got: 'foo'
00:13:33FromDiscord<Elegantbeef> Damn
00:13:53FromDiscord<Elegantbeef> Nim has an implicitly declared `result` variable in procedures
00:14:04FromDiscord<Elegantbeef> So you do not need to make a value that is returned as it already has one
00:14:16FromDiscord<dom96> !eval proc foo(): string = ("hello world"); echo(foo())
00:14:19NimBotCompile failed: /usercode/in.nim(1, 23) Error: expression '"hello world"' is of type 'string' and has to be used (or discarded)
00:14:27FromDiscord<!Patitotective> lmao
00:14:33FromDiscord<Sun「無用」> So that's what makes it less readable...
00:14:35FromDiscord<dom96> ffs Nim
00:16:01FromDiscord<Sun「無用」> sent a code paste, see https://play.nim-lang.org/#ix=45RW
00:16:16FromDiscord<Sun「無用」> And it did nothing.
00:16:21FromDiscord<!Patitotective> proc
00:16:29FromDiscord<!Patitotective> (edit) "useproc ... " added "btw"
00:16:41FromDiscord<Sun「無用」> but it has no side effects though
00:16:46FromDiscord<dom96> It's an IRC bot
00:16:51FromDiscord<dom96> it has no knowledge of Discord's long messages
00:17:39FromDiscord<b1rdf00d> sent a code paste, see https://play.nim-lang.org/#ix=45RX
00:17:42FromDiscord<!Patitotective> 🤯
00:17:46FromDiscord<Sun「無用」> In reply to @b1rdf00d "You can do this": yea, that's more readable to me
00:17:56FromDiscord<ElegantBeef> Damn bridge
00:17:56FromDiscord<Sun「無用」> that result thing is kinda hard to read
00:18:06FromDiscord<ElegantBeef> It's really not
00:18:08FromDiscord<b1rdf00d> In reply to @Avahe "Also <@775600941985431562> your `echo": Ah thanks 😄 now that you mention it _yes_ it should be
00:18:14FromDiscord<ElegantBeef> It's odd but it's a fantastic feature
00:18:26FromDiscord<ElegantBeef> Odin also has named return values
00:18:29FromDiscord<Sun「無用」> In reply to @ElegantBeef "It's really not": it is... for me.
00:18:52FromDiscord<ElegantBeef> you also havent written Nim extensively so it's a new feature that is odd and not ingrained in your toolbag
00:19:06FromDiscord<ElegantBeef> It might be less readable initially i'll give you that
00:19:29FromDiscord<Sun「無用」> In reply to @ElegantBeef "you also havent written": I'm literally writing it for the whole day.
00:19:52FromDiscord<!Patitotective> you haven't written enough
00:20:06FromDiscord<Sun「無用」> I also have already used it b4
00:23:20FromDiscord<b1rdf00d> sent a long message, see http://ix.io/45RY
00:23:57FromDiscord<Elegantbeef> return if you have an optimization that needs it, else use result/implicit
00:24:09FromDiscord<Sun「無用」> I just find hard to know what the value inside result is, most of the time, but ok.
00:24:26FromDiscord<b1rdf00d> In reply to @Elegantbeef "return if you have": what's the penalty of using result instead of return ?
00:24:32FromDiscord<Elegantbeef> How is it any different to `var res = 0` or w/e
00:24:36FromDiscord<Elegantbeef> There is no penalty
00:24:43FromDiscord<Elegantbeef> Or atleast in theory there isnt
00:25:03FromDiscord<Elegantbeef> Nim NVROs the result so it's often actually rewritten as a proc that takes a `result: var T`
00:25:17FromDiscord<b1rdf00d> ah okay, so the optimisation is just depending on whether an early return is faster?
00:25:27FromDiscord<Elegantbeef> Consider a search function
00:25:39FromDiscord<b1rdf00d> yep yep
00:26:03FromDiscord<Sun「無用」> In reply to @Elegantbeef "How is it any": you can f8(or whatever the keybind is) and you can know the default value and the logic used
00:26:15FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45RZ
00:26:24FromDiscord<Elegantbeef> Result like all other nim variables is 0-init'd↵(@Sun「無用」)
00:26:45FromDiscord<Elegantbeef> So it's identical to that declaration
00:26:56FromDiscord<Elegantbeef> if there is no assignment to result it's' `default(typeof(result))`
00:27:15FromDiscord<Sun「無用」> what...
00:27:34FromDiscord<Elegantbeef> The default value of all variables in nim is the 0'd state of memory
00:28:20FromDiscord<Elegantbeef> sent a code paste, see https://paste.rs/1Vd
00:28:27FromDiscord<Elegantbeef> Where is the ambiguity in what result is?
00:28:40FromDiscord<Sun「無用」> in 1 liners? none
00:29:07FromDiscord<Elegantbeef> Well provide a case where an explicit return value is less ambiguous
00:29:08FromDiscord<Sun「無用」> in multiple lines, I just get lost with result
00:29:23FromDiscord<Sun「無用」> sent a code paste, see https://play.nim-lang.org/#ix=45S0
00:29:41FromDiscord<Sun「無用」> for me, at least
00:29:54FromDiscord<Elegantbeef> there is a single for loop so that `break` is logically equivlent to a return
00:30:12FromDiscord<Elegantbeef> so mentally rewrite it as a return like it should've been written 😄
00:30:33FromDiscord<Sun「無用」> which makes it harder to read...
00:30:37FromDiscord<Sun「無用」> literally what I said
00:33:09FromDiscord<b1rdf00d> In reply to @dom96 "even startups can do": can confirm I've used sqlite instead of a "real" database for months at a start up
00:42:37FromDiscord<dom96> Nim's forum uses sqlite too
01:29:47FromDiscord<b1rdf00d> In reply to @Patitotective "use proc btw": I've been trying to use `func` everywhere unless the compiler tells me I can't, is there a reason to prefer `proc` over `func` when `func` compiles?
01:30:12FromDiscord<b1rdf00d> (feels similar to preferring `let` over `var`)
01:30:45FromDiscord<!Patitotective> In reply to @b1rdf00d "I've been trying to": i have the same question
01:48:48FromDiscord<Prestige> I believe it's mostly for the developer, if you have a function in which you intend there to be no side effect
01:48:59*krux02 quit (Remote host closed the connection)
01:58:21FromDiscord<!Patitotective> makes sense
01:59:03FromDiscord<!Patitotective> In reply to @b1rdf00d "(feels similar to preferring": i wonder if func has some optimization
02:01:17FromDiscord<Elegantbeef> In theory it could but it does nothing but mutation analysis
02:01:43FromDiscord<Elegantbeef> though nim already does that optimisation now that i say that
02:21:39NimEventerNew Nimble package! nimtest - Simple testing framework for Nim, see https://github.com/avahe-kellenberger/nimtest
02:34:27FromDiscord<flywind> In reply to @treeform "Do you have a": Thanks, but this is a ORC or nimble bug, not related to flatty.
02:50:14FromDiscord<treeform> ok
02:53:33FromDiscord<ghoom> what's the difference between iterators and coroutines?
03:08:39FromDiscord<ajusa> Is there a way for this to be valid? Or do type aliases not have "priority" when figuring out which proc to call? https://play.nim-lang.org/#ix=45Sn
03:08:59FromDiscord<Elegantbeef> Type aliases are aliases
03:08:59FromDiscord<Elegantbeef> They're not new distinct types
03:09:40FromDiscord<Elegantbeef> You cannot dispatch on an alias that points to the same type as it's the same procedure definition
03:14:18FromDiscord<ajusa> Cool, thanks for explaining! Just wanted to make sure I wasn't missing some workaround.
03:15:41FromDiscord<ajusa> Is there a way to import all procs from a module except for a specific overload of one? Eg, I want to import just integer multiplication but not float multiplication. They share the same name, but different signatures
03:16:48FromDiscord<Elegantbeef> Doubtful though
03:16:51FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45Sq
03:17:01FromDiscord<Elegantbeef> Guess it'd be `proc(a, b: int): int)`
03:21:41FromDiscord<ajusa> so it'd be `import math except proc(a, b: int): int` or something like that? I recall trying to do this a few months ago and couldn't figure out the syntax lol
03:23:13FromDiscord<Elegantbeef> If it worrks
03:23:18FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45Sr
03:24:00FromDiscord<Elegantbeef> It will not work
03:24:07FromDiscord<Elegantbeef> Import explicitly looks for an identifier
03:25:19FromDiscord<ajusa> I'm trying to import everything except for a proc
03:25:37FromDiscord<Elegantbeef> Yea you cannot do it for a specific overload
03:25:38FromDiscord<ajusa> but I see your point, import and export look for an identifier
03:25:51FromDiscord<ajusa> (edit) "export" => "export/except"
03:36:31*arkurious quit (Quit: Leaving)
04:24:29*pch joined #nim
04:54:58FromDiscord<Abigail> Good evening, I have been having some trouble converting some C++ to Nim ( As a way to learn it ) and I am running into some errors that I haven't been able to figure out. The closest thing I have found via google was something dealing with templates so it did not really help out. Here is the code and the answers if anyone is able to help out: https://play.nim-lang.org/#ix=45SE
04:57:40FromDiscord<Rika> Put discard at the start of the line
04:57:57FromDiscord<Rika> All procedures that return a value must have a consumer in some way, be it echo or discard
04:58:09FromDiscord<Elegantbeef> Nim does not allow unhandled values
04:58:16FromDiscord<Rika> You cannot bare call procedures with returns like that
05:08:45FromDiscord<Abigail> In reply to @Rika "All procedures that return": so, if i am understanding correctly, discard needs to be put at the before the call since the output is not being assigned to a variable nor being echoed?
05:09:15FromDiscord<Elegantbeef> Yes
05:09:30FromDiscord<Elegantbeef> The procedure returns a value and you do not handle it as such you need to explicitly discard it
05:09:52FromDiscord<Abigail> okay that makes some sense, def different than all the other langauges i have touched so far.
05:10:20FromDiscord<Elegantbeef> This might seem odd but Nim has very nice expressions and definite usage makes those nicer
05:10:35FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45SI
05:11:04FromDiscord<Elegantbeef> You're less likely to use a procedure that you want the value in something like the above
05:11:57FromDiscord<Abigail> oooh
05:12:22FromDiscord<Abigail> seems kind of like that one thing in python i cannot remember the name of right now
05:12:38FromDiscord<Rika> No it’s not like the walrus operate
05:12:40FromDiscord<Rika> (edit) "operate" => "operator"
05:14:21FromDiscord<huantian> python does ternaries similiarly ish to nim tho
05:14:21FromDiscord<Abigail> so not like list comprehension?
05:14:51FromDiscord<Abigail> and have yet to use the walrus operator in python so far, most of what i have used is just back end stuff with flask and a bot with discordpy
05:15:06FromDiscord<Rika> This isn’t like comprehensions
05:15:17FromDiscord<Rika> This is just if as an expression
05:21:37FromDiscord<huantian> sent a code paste, see https://play.nim-lang.org/#ix=45SL
05:21:58FromDiscord<Abigail> okay that makes a bit of sense
05:32:54FromDiscord<Abigail> finally got the program working but it did bring about one question - is there a way to have a mutable default value for variable? Didn't see anything on the forums or SO, and didnt see anything that connected the two on the website. This is what I have it at now, thankfully at a runnable state https://play.nim-lang.org/#ix=45SO
05:33:31FromDiscord<Elegantbeef> Make an overload is the way
05:33:51FromDiscord<Elegantbeef> A default mutable doesnt make sense in a Nim context
05:34:26FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45SP
05:34:29FromDiscord<Elegantbeef> Or w/e you want `count` to be
05:34:50FromDiscord<Abigail> oooh - that is neat
05:35:59FromDiscord<#!/rip/luke> Can I layer 2 images in pixie
05:36:23FromDiscord<Elegantbeef> Of course
05:57:48FromDiscord<Girvo> sent a code paste, see https://play.nim-lang.org/#ix=45ST
05:58:00FromDiscord<Elegantbeef> And i'd hate you for it so that's all fine
05:58:10FromDiscord<Girvo> 😘
05:58:11FromDiscord<Elegantbeef> Vertical code is always better than horizontal and i wont hear otherwise
06:21:53NimEventerNew Nimble package! jitter - A git-based binary manager for linux., see https://github.com/sharpcdf/jitter
06:23:32FromDiscord<Girvo> Is there a compile-time way to ensure a template or proc is only ever called once? just curious
06:24:50FromDiscord<!!sharpcdf!!> In reply to @NimEventer "New Nimble package! jitter": may or may not be fully a fully functional installation yet lol
06:25:56FromDiscord<Elegantbeef> Proc no, template yes↵(@Girvo)
06:26:39FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45SY
06:27:11FromDiscord<Girvo> Oh neat
06:28:35FromDiscord<Elegantbeef> Whoops i leftout the `myChecker = true` from it
06:28:39FromDiscord<Elegantbeef> But you can reason that one 😄
06:29:47FromDiscord<Elegantbeef> it'd of course be in a static context
06:30:09FromDiscord<Elegantbeef> You could for a procedure wrap the user side procedure with a template like the above
06:32:10FromDiscord<Elegantbeef> That'd ensure a single callsite
06:32:41FromDiscord<Girvo> Yeah. I really was only considering it for templates anyway -- theres some global referenc-y stuff around FreeRTOS message queue handles that `app_main` needs to create, but I wanted to keep the actual creation code in the same module as the thread code itself
06:32:42FromDiscord<Elegantbeef> You then could use something like `once` and have an assertion if that procedure is ran more than once
06:32:59FromDiscord<Girvo> But it really can't be called more than once lol. Not really an ideal use of templates, but hey
06:35:31FromDiscord<Elegantbeef> This is where i say you could make a macro to emit the template for you 😛
06:35:43FromDiscord<Elegantbeef> Then no one sees the template and you hide the dirt
06:36:06FromDiscord<Elegantbeef> Though you'd need a big rug or a big brush to hide it under
06:37:53FromDiscord<Girvo> And tell people to _never ever_ open those files mwahaha
06:44:30FromDiscord<b1rdf00d> In reply to @b1rdf00d "I'm working through ray": bumping this, I'd assumed leaving inlining up to the compiler was the was to go?
06:45:14FromDiscord<Elegantbeef> You annotate procedures as inline if you want to coax the compiler to do it
06:49:29FromDiscord<b1rdf00d> I thought the compiler would be clever enough to work this one out though?
06:50:31FromDiscord<Elegantbeef> They're not always as smart as you'd want them to
06:50:40FromDiscord<Elegantbeef> It was a release with lto right?
06:51:05FromDiscord<b1rdf00d> yeah like this: `nim c -d:release -d:lto main.nim`
06:51:22FromDiscord<b1rdf00d> (I defined lto in my config.nims)
06:52:38FromDiscord<Girvo> Sufficiently-smart compilers sadly aren't quite here yet, in my opinion. Getting closer, mind you.
06:54:59FromDiscord<b1rdf00d> 👍 yep cool, so build up experience with the compiler to get a feel for when to nudge it sounds like the way to go
06:55:17FromDiscord<Girvo> There is likely some particular construct or approach you can use to get it to infer it should be inlined, but honestly its simpler to just tell it 🙂
06:56:09FromDiscord<b1rdf00d> I think in this case just flat out saying to inline was the right approach
06:58:02FromDiscord<Elegantbeef> So Nim has 2 mechanisms in regard to inlining, it has the inline pragma for saying "Hey compiler if you've got nothing better to do feel free to inline" and then templates/macros which enable yo to say "I've got this inline is just better"
06:58:20FromDiscord<Elegantbeef> Based off what mratsim the inline pragma is sufficient in most cases
06:59:56FromDiscord<Girvo> Though templates are super useful in my case where I can't trust the horrific C compiler ESP-IDF uses under the hood, and I _really_ want to make sure certain blocks are inlined rather than actual functions
07:00:18FromDiscord<Girvo> I'm not even sure whatever god-awful system theyre using can do LTO lol
07:00:36FromDiscord<Elegantbeef> Girvo if you're going to insist single lined if expressions you can atleast just leave it at "sufficient in most cases" 😛
07:00:55FromDiscord<Girvo> 😄 😄 😄
07:01:28FromDiscord<Girvo> Note to self: send a heap of PRs to beef's repos single-lining all expressions
07:01:49FromDiscord<Elegantbeef> Just make all my procedures one lined with expressions
07:04:57FromDiscord<#!/rip/luke> what optimizations can i make, most of the code is in src/dye.nim and lib/colors.nim
07:05:05FromDiscord<#!/rip/luke> https://github.com/Infinitybeond1/dye
07:05:14FromDiscord<Elegantbeef> Time to use a profiler
07:05:37FromDiscord<#!/rip/luke> wym
07:05:41FromDiscord<Rika> Profile your code
07:05:45FromDiscord<Rika> See what is slow
07:05:57FromDiscord<Elegantbeef> Jesus christ luke what did i say about if i seen you use `split` to parse code?!
07:06:21FromDiscord<Elegantbeef> I'm going to get matthew mark and john involved if you keep this up
07:06:35FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Jesus christ luke what": what line
07:06:49FromDiscord<#!/rip/luke> ohh the csv
07:06:59FromDiscord<#!/rip/luke> whats wrong with that
07:07:12FromDiscord<Rika> You ask for optimisations
07:07:19FromDiscord<Rika> Yet you ask what’s wrong with split
07:07:27FromDiscord<#!/rip/luke> wouldnt something like std/parsecsv slower?
07:07:30FromDiscord<Elegantbeef> To be fair rika it's not in the hot path
07:07:32FromDiscord<Rika> Why
07:07:37FromDiscord<Rika> Why would that be slower
07:07:38FromDiscord<#!/rip/luke> (edit) "wouldnt something like std/parsecsv ... slower?" added "be"
07:07:51FromDiscord<Rika> I don’t actually know why you think it would be slowe
07:07:54FromDiscord<Rika> (edit) "slowe" => "slower"
07:07:56FromDiscord<#!/rip/luke> idk its bigger?
07:08:08FromDiscord<Elegantbeef> Code size does not indicate speed
07:08:10FromDiscord<Rika> Sounds irrational to me
07:08:30FromDiscord<#!/rip/luke> oh well ok ill use parsecsv
07:08:34FromDiscord<Elegantbeef> Split is slow cause it allocates a sequence of strings which means it uses a shit ton of memory and allocations
07:09:05FromDiscord<#!/rip/luke> oh
07:09:25FromDiscord<Rika> In reply to @Elegantbeef "Code size does not": What’s slower, 99 discards or 1 “allocate 15 terabytes of memory”
07:09:40FromDiscord<Elegantbeef> Lol
07:09:48FromDiscord<Elegantbeef> Anyway profile your code
07:09:56FromDiscord<Elegantbeef> It's the easiest way to see where to optimise
07:10:07FromDiscord<Elegantbeef> Unless you suddenly develop the knowledge of a more seasoned developer
07:10:19FromDiscord<Rika> In which case he’d still profile his code
07:10:27FromDiscord<Elegantbeef> Not always
07:10:46FromDiscord<Rika> How would you know 😛
07:10:54FromDiscord<Prestige> Heh
07:11:01FromDiscord<Elegantbeef> Like looking at patitos code for their text editor took me all of 10 seconds to see that he was using `runeLen` multiple times per frame
07:11:10FromDiscord<Elegantbeef> No one said a seasoned professional developer
07:11:13FromDiscord<Rika> 9 seconds too slow
07:11:14FromDiscord<Elegantbeef> I'm seasoned like a good fish
07:11:17FromDiscord<Prestige> Profiling is too good
07:11:39FromDiscord<#!/rip/luke> In reply to @Elegantbeef "I'm seasoned like a": 😳
07:11:55FromDiscord<#!/rip/luke> i would think that youre seasoned like a steak
07:12:03FromDiscord<Elegantbeef> Nah fish \> steak
07:12:28FromDiscord<Prestige> Steak is superior
07:12:31FromDiscord<#!/rip/luke> Elegantfish > Elegantbeef
07:12:40FromDiscord<Prestige> Lol
07:12:42FromDiscord<Elegantbeef> Prestige it's ok to be wrong just dont be it near me
07:13:02FromDiscord<Girvo> In reply to @Elegantbeef "Unless you suddenly develop": Tbh knowledge can lead you astray too. Often things can break your intuition when it comes to performance, even with years and years of experience.
07:13:13FromDiscord<Elegantbeef> It's not a performance related thing but please never use `getCurrentException`
07:13:13FromDiscord<Girvo> So, profile profile profile
07:13:25FromDiscord<Elegantbeef> Depends on what it is girvo
07:13:34FromDiscord<Rika> I’m gonna rename myself to elegant fish
07:13:37FromDiscord<Prestige> Red meat too good
07:13:46FromDiscord<Prestige> I will eat a steak in your honor
07:13:48FromDiscord<Elegantbeef> Having 800 calls to runelen in a draw loop of program is comical 😄
07:14:05FromDiscord<Elegantbeef> But yes profiling is good
07:14:06FromDiscord<#!/rip/luke> i dont think i can use parsecsv for my usecase
07:14:09FromDiscord<Elegantbeef> But also just having a brain is better
07:14:18FromDiscord<Elegantbeef> No you can use `parseUtils` though
07:14:30FromDiscord<Prestige> Profiling > logic
07:14:32FromDiscord<Girvo> `std/pegs` ain't that fast is it hey?
07:14:46FromDiscord<Elegantbeef> No clue i'd use npeg before std/pegs
07:14:50FromDiscord<Girvo> Logic gets you all the simple fixes for free. Profiling gets you the rest of the way there 🙂
07:14:52FromDiscord<Elegantbeef> And no clue how fast npeg is
07:14:59FromDiscord<#!/rip/luke> also i dont think the split really affects the performance as its only ran once and isnt in the loop\
07:15:08FromDiscord<Elegantbeef> It doesnt but i said do not use it
07:15:23FromDiscord<#!/rip/luke> but parsecsv isnt as flxible
07:15:27FromDiscord<Elegantbeef> ....
07:15:54FromDiscord<#!/rip/luke> like lets say theres multiple lines, it wont wok
07:15:57FromDiscord<#!/rip/luke> (edit) "wok" => "work"
07:16:02FromDiscord<Elegantbeef> Like i said parseutils exists
07:16:32FromDiscord<#!/rip/luke> never heard of it
07:16:53FromDiscord<Elegantbeef> `file.parseUntil(',')` goes brr
07:16:59FromDiscord<Girvo> https://nim-lang.org/docs/parseutils.html
07:17:15FromDiscord<Girvo> The standard library has a heap of cool toys 🙂
07:18:05FromDiscord<Rika> Car battery included
07:18:18FromDiscord<Girvo> Plus some jumper cables
07:18:54FromDiscord<#!/rip/luke> In reply to @Elegantbeef "`file.parseUntil(',')` goes brr": that returns an in lol\
07:19:03FromDiscord<#!/rip/luke> (edit) "in" => "int"
07:19:22FromDiscord<Girvo> Yep, as it says
07:19:28FromDiscord<Girvo> > Parses a token and stores it in token. Returns the number of the parsed characters or 0 in case of an error
07:20:17FromDiscord<#!/rip/luke> yea but it wont work for my usecase lmao
07:20:21FromDiscord<Rika> Why
07:21:34FromDiscord<#!/rip/luke> because im trying to parse some sorta csv, i want the values not the number of charecters
07:21:46FromDiscord<Elegantbeef> jesus christ
07:21:52FromDiscord<Elegantbeef> Ok so i'll write your parser
07:22:16FromDiscord<#!/rip/luke> why tho? thats just overengineering
07:22:33FromDiscord<#!/rip/luke> if split gets the job done in a timely fashion, whats wrong with it
07:22:45FromDiscord<Elegantbeef> It doesnt
07:22:49FromDiscord<Elegantbeef> Cause you pass it to `rmTag`
07:23:08FromDiscord<#!/rip/luke> yea
07:23:42FromDiscord<#!/rip/luke> but the longest parts are the flip and luma functions, i dont know whats wrong with them
07:23:52FromDiscord<Elegantbeef> Profile them
07:24:08FromDiscord<#!/rip/luke> what should i use
07:24:26FromDiscord<#!/rip/luke> is there something in std?
07:24:27FromDiscord<Elegantbeef> `valgrind --tool=callgrind`
07:24:28FromDiscord<Elegantbeef> kcachegrind
07:24:32FromDiscord<Elegantbeef> perf also exists
07:24:57FromDiscord<b1rdf00d> the in built nim one is pretty handy for not needing to install anything else (in my limited experience)
07:24:58FromDiscord<#!/rip/luke> In reply to @Elegantbeef "`valgrind --tool=callgrind`": ok so its a cli thing
07:24:58FromDiscord<Girvo> Seconding `valgrind` -- its _always_ worth setting up and using on any project 🙂 will save you lots of pain if performance is the goal!
07:25:24FromDiscord<Girvo> Valkyrie also exists as a wrapper around valgrind's CLI
07:25:28FromDiscord<Girvo> If you really need a gui
07:25:39FromDiscord<b1rdf00d> but yeah valgrind would be the ultimate way to go
07:25:42FromDiscord<#!/rip/luke> nah i can use a cli
07:25:59FromDiscord<Girvo> https://valgrind.org/downloads/guis.html theres a few others listed there too. It can be handy for pulling apart call graphs, to have a gui, but for basic usage the CLI is great
07:26:23FromDiscord<vindaar> it depends, valgrind really impacts performance heavily. use perf if that is an issue for near native performance
07:26:55FromDiscord<Girvo> vindaar, if the instrumentation overhead is a pain, ASAN is really good
07:27:16FromDiscord<Girvo> specifically around memory error stuff anyway
07:27:59FromDiscord<Girvo> `cachegrind` is invaluable though IMHO in terms of really nailing hot-path code perf
07:28:34FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45TC
07:28:36FromDiscord<Elegantbeef> That's all that's required here
07:28:47FromDiscord<vindaar> interesting. guess I have reading to do↵(@Girvo)
07:28:59FromDiscord<Elegantbeef> Could even do a `skipUntil` instead of `skipWhile`
07:30:21FromDiscord<#!/rip/luke> oh
07:30:41FromDiscord<Girvo> In reply to @vindaar "interesting. guess I have": Between Valgrind's toolset and ASAN for really performant memory error checking, I basically never need anything else haha
07:31:37FromDiscord<Elegantbeef> I know leo says that perf is pretty good in the overhead department
07:32:07FromDiscord<Girvo> Yeah it's decent. I think I'm just biased coz I've been using valgrind since the late 2000s at this point lol
07:33:37FromDiscord<jseb> hello, i'm waiting for the book « mastering nim », but is it true there's a problem with the first sample of the book ?
07:35:15FromDiscord<jseb> (edit) "?" => "?↵(this comment of the book: https://www.amazon.fr/product-reviews/B0B4R7B9YX/ref=acr_search_hist_1?ie=UTF8&filterByStar=one_star )"
07:35:47FromDiscord<Elegantbeef> Araq uses a m1 to develop afaik so i'm uncertain if that's true
07:36:05FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Anyway to make the": wow its faster now
07:36:17FromDiscord<Elegantbeef> That's sarcasm
07:36:18FromDiscord<#!/rip/luke> thx
07:36:19FromDiscord<Elegantbeef> But anyway
07:36:30FromDiscord<#!/rip/luke> no its actually faster
07:36:56FromDiscord<Girvo> Hah is Araq on mac these days?
07:37:04FromDiscord<Elegantbeef> I believe he is
07:37:07FromDiscord<#!/rip/luke> In reply to @ripluke "no its actually faster": 35ish milliseconds faster
07:37:22FromDiscord<Girvo> I remember wayyyyy back when I was one of the few weirdos who was trying to use Nim on a mac when it was quite Windows focused
07:37:37FromDiscord<Elegantbeef> Which is kinda funny given that the defacto versioning program doesnt ship native binaries for mac
07:37:38FromDiscord<#!/rip/luke> https://media.discordapp.net/attachments/371759389889003532/1002117612735574016/unknown.png
07:37:40FromDiscord<Girvo> I think I fixed a bug in the cURL bindings with mac lol
07:39:17FromDiscord<Elegantbeef> I mean it was going to be faster but i do not get how it'ss that much faster given it's not even the hotpath, but this proves profiling matters 😄
07:41:53FromDiscord<Elegantbeef> We'll just go with "Ugly code is generally slow code"
07:42:10FromDiscord<Girvo> In a well designed language, thats typically true I find 🙂
07:42:42FromDiscord<Elegantbeef> Also luke you're still copying image
07:42:49FromDiscord<Elegantbeef> If you want a fast program you dont copy unless you need to
07:44:01FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45TH
07:44:24FromDiscord<Elegantbeef> Not a performance thing again, but cmon you've written the same code more than once, that's a sign it needs to be repeatable
07:44:52FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Not a performance thing": ok
07:46:05FromDiscord<Girvo> Code duplication is not always the enemy, and I often repeat stuff too, initially. But it's always a sign I should go back and examine if it should be a proc (or other construct) instead 🙂
07:46:19FromDiscord<Elegantbeef> See i dislike writing code
07:46:29FromDiscord<Elegantbeef> I hate anything tedious and repetitive
07:46:43FromDiscord<Elegantbeef> So talking to you girvo draws me up the wall 😛
07:46:47FromDiscord<Elegantbeef> drives
07:46:56FromDiscord<Elegantbeef> Fuck not a good insult if you have to correct it
07:47:37FromDiscord<Elegantbeef> Also luke you have the exact same procedure twice just to show a progress bar
07:48:04FromDiscord<Elegantbeef> Remove the overhead branch then just put if checks for the ar logic
07:48:08FromDiscord<Elegantbeef> bar\
07:48:26FromDiscord<Girvo> hahaha yeah would've been a great insult if you stuck the landing mate 😉
07:49:12FromDiscord<Elegantbeef> Really luke i'd say make a single procedure which takes a `proc(color: Color): Color` so you dont have redundant code
07:49:43FromDiscord<Elegantbeef> Might slow the program down a bit though
07:50:08FromDiscord<Elegantbeef> So maybe not ideal
07:50:21FromDiscord<Elegantbeef> Yea girvo i'm fantastic and almost insults
07:50:35FromDiscord<Elegantbeef> Unless it's self deprecation
07:51:00FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Remove the overhead branch": wouldnt thos if checks take longer
07:51:15FromDiscord<Elegantbeef> It's a if bool check that's nothing
07:51:22FromDiscord<#!/rip/luke> ok
07:52:29FromDiscord<#!/rip/luke> In reply to @Elegantbeef "It's a if bool": but i would have to do it every pixel
07:53:04FromDiscord<Elegantbeef> Dont update the bar every pixel?
07:53:14FromDiscord<Elegantbeef> I dont even get the bar
07:53:26FromDiscord<Elegantbeef> Even a complex application should be a few seconds at most
07:53:28FromDiscord<#!/rip/luke> In reply to @Elegantbeef "I dont even get": its opt-in
07:53:41FromDiscord<Elegantbeef> I'm saying i dont even understand it
07:53:53FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Even a complex application": if youre converting an 8k image its kiinda usefule
07:54:10FromDiscord<Elegantbeef> Then do it every line instead of every pixel
07:54:24FromDiscord<Elegantbeef> Doing it every pixel is needlessly slowing the program down as it is
07:54:25FromDiscord<Girvo> Yep. Chunk it, so you don't have to run it constantly 🙂
07:54:35FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Then do it every": thats still super intensive
07:54:42FromDiscord<Elegantbeef> No it's not
07:54:50FromDiscord<Elegantbeef> 8k is only 8000 lines
07:54:52FromDiscord<#!/rip/luke> the duplicated code isnt that bad
07:55:00FromDiscord<Elegantbeef> Duplicated code is terrible
07:55:05FromDiscord<Elegantbeef> Source\: me
07:55:13FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Duplicated code is terrible": the end user wont see it tho
07:55:28FromDiscord<#!/rip/luke> so at the end of the day it doesnt really matter
07:55:31FromDiscord<Elegantbeef> You're right but the most important person that sees the code does
07:56:13FromDiscord<#!/rip/luke> neovim has this nice feature where it collapses codeblocks you arent editing
07:56:18FromDiscord<#!/rip/luke> so it isnt that bad
07:57:19FromDiscord<#!/rip/luke> very useful https://media.discordapp.net/attachments/371759389889003532/1002122568985235456/unknown.png
07:57:34FromDiscord<Elegantbeef> That's not an argument for it
07:57:37FromDiscord<Elegantbeef> But ok you do you
07:57:43FromDiscord<Elegantbeef> I cannot force you to write cleaner code
07:58:09FromDiscord<#!/rip/luke> In reply to @Elegantbeef "I cannot force you": cleaner code at the cost of efficiency
07:58:21FromDiscord<Elegantbeef> It's really not as slow as you think it is
07:58:30FromDiscord<#!/rip/luke> im all for clean code, but id rather have performance
07:58:43FromDiscord<#!/rip/luke> In reply to @Elegantbeef "It's really not as": ok ill benchmark it
08:00:51FromDiscord<#!/rip/luke> hmm for another thing, does nim have any form of a dithering library
08:01:01FromDiscord<Girvo> Profile profile profile! Clean code + optimisations is nearly _always_ faster in my experience 🙂
08:01:37FromDiscord<Girvo> Any particular dithering algorithm you're looking for?
08:01:37FromDiscord<Elegantbeef> This is 100% slower, no clue how much but it's a much nicer way of handling it
08:01:39FromDiscord<Elegantbeef> https://play.nim-lang.org/#ix=45TP
08:01:59FromDiscord<Elegantbeef> If you wanted it to be on par you could make it a template 😄
08:02:17FromDiscord<#!/rip/luke> In reply to @Girvo "Any particular dithering algorithm": nope but floyd-stienburg would be preferable
08:03:30FromDiscord<Elegantbeef> https://play.nim-lang.org/#ix=45TS this is how you'd do the template version
08:04:09FromDiscord<Girvo> I'd likely implement it myself, but failing that I'm sure there's a C implementation that would be easy enough to bind too. Or use shaders, though I'm guessing that wouldn't suit your use-case quite as well
08:04:20FromDiscord<Elegantbeef> sent a code paste, see https://play.nim-lang.org/#ix=45TT
08:05:00FromDiscord<#!/rip/luke> oh
08:05:16FromDiscord<Elegantbeef> And now you have way less redundant code but still performance
08:05:19FromDiscord<#!/rip/luke> In reply to @Girvo "I'd likely implement it": hmm ok ill look for a c one
08:05:27FromDiscord<#!/rip/luke> In reply to @Elegantbeef "And now you have": oh
08:05:44FromDiscord<#!/rip/luke> In reply to @ripluke "hmm ok ill look": i might wrap soko;l
08:05:47FromDiscord<#!/rip/luke> (edit) "soko;l" => "sokol"
08:06:07FromDiscord<Girvo> Oh yeah nice
08:06:32FromDiscord<Girvo> Oh it has Nim bindings already haha
08:06:40FromDiscord<Girvo> https://github.com/floooh/sokol-nim
08:06:53FromDiscord<Girvo> I'd use those if it has the algorithm you're after!
08:08:40FromDiscord<Girvo> Otherwise: https://github.com/xavierpinho/image-dithering-demo/tree/main/src you'd really only need to bind `image.h` and `dither.h` from what I can tell, and they're both very simple. Though no promises on how well implemented they are, I was just looking haha
08:09:05FromDiscord<#!/rip/luke> In reply to @Girvo "https://github.com/floooh/sokol-nim": O.o thats saves some work lol
08:09:27FromDiscord<Girvo> Haha the moment I saw a `bindgen` folder I went looking. That usually means someones generated Nim bindings already 😉
08:10:31FromDiscord<Girvo> https://www.shadertoy.com/view/Xst3W7 but its also a perfect application for a shader, too, if you go down that path later on!
08:11:26FromDiscord<Elegantbeef> Yea they're basically doing CPU shaders 😄
08:15:12FromDiscord<Rika> is there a reason you want floyd steinberg
08:15:19FromDiscord<Rika> i dont think FS is shaderisable
08:15:27FromDiscord<Rika> since its a convolution alg
08:15:43FromDiscord<#!/rip/luke> lol i dont want shaders
08:16:14FromDiscord<#!/rip/luke> i want the ditherng so the converted image looks better with even only a few colors
08:16:36FromDiscord<#!/rip/luke> (edit) "better" => "great"
08:18:51FromDiscord<Elegantbeef> "I dont want shaders"↵Is literally doing pixel shaders on the cpu
08:18:57FromDiscord<#!/rip/luke> In reply to @Girvo "Otherwise: https://github.com/xavierpinho/image-dit": lol yea im gonna use this as its alot more simple than sokol and it has a variety of algos
08:19:16FromDiscord<#!/rip/luke> In reply to @Elegantbeef ""I dont want shaders"": wot? where do i have shaders
08:19:27FromDiscord<Elegantbeef> you literally have a pixel shader
08:19:39FromDiscord<Elegantbeef> A pixel shader renders once per every rendered pixel
08:19:50FromDiscord<Elegantbeef> You've got a CPU based pixel shader, congrats!
08:20:00FromDiscord<#!/rip/luke> oh is that good or bad?
08:20:10FromDiscord<Elegantbeef> Ask your CPU how it likes it
08:20:17FromDiscord<#!/rip/luke> ok
08:20:37FromDiscord<#!/rip/luke> idk
08:20:48FromDiscord<#!/rip/luke> my cpu didnt answer
08:21:16FromDiscord<Rika> i think you should look into the a-dither algorithm i sent above, its easily parallelisable into a shader
08:21:24FromDiscord<Rika> i dont think FS is
08:21:30FromDiscord<creikey> is there any pure nim libraries out there doing skeletal animation with opengl?
08:21:42FromDiscord<#!/rip/luke> In reply to @Rika "i think you should": where did u send it?
08:22:06FromDiscord<Rika> In reply to @Rika "https://pippin.gimp.org/a_dither/": .
08:22:09FromDiscord<Elegantbeef> https://github.com/guzba/gltfviewer i think this does
08:22:16FromDiscord<Elegantbeef> It's not a library but it's a reference
08:22:20FromDiscord<#!/rip/luke> ok
08:23:42FromDiscord<Rika> no error diffusion algorithm is easily parallelisable because each pixel depends on all pixels before it sooooooooo
08:23:52FromDiscord<Rika> ordered dithering is a must
08:24:20FromDiscord<#!/rip/luke> ok
08:24:47FromDiscord<Rika> i have to go
08:25:05FromDiscord<#!/rip/luke> if i wanted to wrap this should i use c2nim or futhark https://github.com/xavierpinho/image-dithering-demo/blob/main/src/image.h
08:25:29FromDiscord<Elegantbeef> It's like 10 functions you can probably do it by hand 😄
08:26:07FromDiscord<#!/rip/luke> its my first time 💀
08:26:08FromDiscord<Elegantbeef> But both will work so you do you
08:28:59FromDiscord<#!/rip/luke> wait with c2nim the functions have no body lol
08:29:09FromDiscord<#!/rip/luke> do i have to do it myself
08:36:35*vicecea quit (Remote host closed the connection)
08:37:05*vicecea joined #nim
08:45:15*rockcavera quit (Remote host closed the connection)
08:48:19FromDiscord<Elegantbeef> Imported code doesnt have bodies
08:49:28FromDiscord<creikey> In reply to @Elegantbeef "https://github.com/guzba/gltfviewer i think this": I don't think this does skeletal animation?
09:15:06FromDiscord<#!/rip/luke> In reply to @Elegantbeef "Imported code doesnt have": Wym
09:15:21FromDiscord<#!/rip/luke> So would the functions work?
09:15:32FromDiscord<#!/rip/luke> Or do I make the bodies
09:15:48FromDiscord<Elegantbeef> The code is imported from C
09:18:46FromDiscord<#!/rip/luke> Wait so can I use the functions tho
09:18:56FromDiscord<#!/rip/luke> Like would everything work?
09:19:08FromDiscord<Elegantbeef> I'm too tired to explain this nicely
09:19:16FromDiscord<Elegantbeef> Someone else want to step in 😄
09:24:49FromDiscord<creikey> In reply to @ripluke "Or do I make": imagine if this is how it was
09:24:59FromDiscord<creikey> like to use a C library you have to also write the implementation in nim
09:25:17FromDiscord<#!/rip/luke> You don't?
09:25:39FromDiscord<bren> The functions are just declarations. Their definitions are still in the C code that gets compiled into a library which you’ll link to your nim code
09:25:40FromDiscord<#!/rip/luke> Lmao I'm very new to this wrapping thing
09:26:13FromDiscord<creikey> In reply to @ripluke "You don't?": that would be such a bananas amount of work
09:26:13FromDiscord<#!/rip/luke> In reply to @bren "The functions are just": Ohh so the Nim code just exposes the functions/types
09:26:17FromDiscord<creikey> like why even have the C library
09:26:24FromDiscord<creikey> if you have to rewrite it in nim
09:27:04FromDiscord<#!/rip/luke> In reply to @creikey "like why even have": Maybe as a starting point? Idk
09:27:11FromDiscord<#!/rip/luke> Seems kinda dumb now
09:27:51FromDiscord<#!/rip/luke> Do I have to do anything for the c code to be compiled?
09:27:59FromDiscord<fbpyr> I think the demo chapter in `nim in action` about c ffi explains that nicely\:↵https://livebook.manning.com/book/nim-in-action/chapter-8/
09:28:03FromDiscord<creikey> In reply to @ripluke "Do I have to": yeah it's weird, check out the nim in action
09:28:06FromDiscord<creikey> yeah that chapter is really good
09:28:08FromDiscord<#!/rip/luke> Oh
09:28:11FromDiscord<#!/rip/luke> Ok
10:12:52FromDiscord<dom96> Thanks for the kind words and linking to my book :)
10:13:53FromDiscord<dom96> You can get a PDF of this chapter here: https://manning-content.s3.amazonaws.com/download/5/08b3cf8-cea3-42c1-97b5-b5047c472b73/SampleCh08.pdf
10:14:04FromDiscord<dom96> (With higher resolution images and generally nicer formatting IMO)
10:14:38FromDiscord<flywind> sent a code paste, see https://play.nim-lang.org/#ix=45Uu
10:16:40FromDiscord<dom96> ORC doesn't support that?
10:18:38FromDiscord<flywind> ORC doesn't like global pragmas, som issues reported => https://github.com/nim-lang/Nim/labels/%7B.global.%7D
10:18:46FromDiscord<flywind> (edit) "som" => "some"
10:20:14FromDiscord<flywind> Lift it to a global scopes directly, it works.
10:20:23FromDiscord<flywind> (edit) "scopes" => "scope"
10:20:31FromDiscord<flywind> (edit) removed "a" | "scope" => "scopes"
10:21:04FromDiscord<flywind> (edit) "Lift" => "Lifted" | "directly," => "directly and"
10:22:48FromDiscord<dom96> this is emulating memoization afaik, might just be better to do that properly instead of (ab)using {.global.}
10:24:47FromDiscord<flywind> I see, thanks! Btw, is it ready to bump the nimble hash for Nim? Otherwise we need to fix the ORC bug instead.
10:27:40FromDiscord<dom96> That I do not know myself. Status folks might know better. But worst case we can always revert anyway so go for it :)
10:28:03FromDiscord<dom96> I still haven't given the new Nimble a proper test myself
10:28:33FromDiscord<dom96> maybe you can give it a go and see if you run into any problems?
10:29:06FromDiscord<flywind> Thanksa lot, I will make a PR to bump the hash.
10:29:11FromDiscord<flywind> In reply to @dom96 "maybe you can give": Ok.
10:43:18*jmdaemon quit (Ping timeout: 264 seconds)
11:02:27FromDiscord<d4rckh> whats the correct syntax for adding an "else if" to the when block?
11:02:56FromDiscord<d4rckh> sent a code paste, see https://play.nim-lang.org/#ix=45UW
11:03:02FromDiscord<d4rckh> (edit) "https://play.nim-lang.org/#ix=45UW" => "https://play.nim-lang.org/#ix=45UX"
11:10:35FromDiscord<Rika> `elif`
11:12:26FromDiscord<d4rckh> ty
11:35:57FromDiscord<LaughingBubba> Hi all, making my way through the Nim in Action book and I'm curious as to why `msg.username = $jsonData["username"]` passes the compile stage (assignment source == target type) but then fails the assert? PS: I thought the $ would have cast it as a string ...
11:42:15FromDiscord<planetis> What assert? If "username" is not found it raises KeyError
11:43:01FromDiscord<planetis> $ converts to string a Jsonnode in this case
11:50:52*spff left #nim (#nim)
11:52:14NimEventerNew Nimble package! trayx - Ray tracing, see https://github.com/teob97/T-RayX
11:55:32FromDiscord<leetnewb> cosmopolitan
11:56:51FromDiscord<sOkam!> I'm reading about `arc/orc`, and I keep finding references to some `cycle collector`, but I can't find anything on google that explains what a cycle collector is (at least with the keywords I searched)↵Can you point me into some keywords / article / search to understand what that collector is, and therefore be able to follow the explanation? 🤔
11:57:45FromDiscord<b1rdf00d> https://nim-lang.org/blog/2020/10/15/introduction-to-arc-orc-in-nim.html ?
11:59:16FromDiscord<b1rdf00d> My understanding is it's a garbage collector that specifically collects cyclical data structures (e.g. like a graph)
11:59:22FromDiscord<sOkam!> oh, i read that article. must have skimmed through those two paragraphs where it gives an idea of what they are 🤔
12:00:26FromDiscord<b1rdf00d> I've run into an issue before shared_ptr in c++ where you have objects referencing each other so nothing is ever freed. But if the graph isn't needed anymore then the whole thing should be freed
12:00:42FromDiscord<flywind> ORC uses trial deletion to collect cycles.
12:00:53FromDiscord<sOkam!> Are all those calls to \`=destroy\`, \`=sink\`, etc... required to be used manually? or are they just explaining what they are and how they work in the background, but they are automatic and the code is the same as with refc?
12:01:39FromDiscord<sOkam!> In reply to @b1rdf00d "I've run into an": oh, that explains it! ty, makes sense
12:02:17FromDiscord<flywind> In reply to @sOkam! "I'm reading about `arc/orc`,": It implements the first algorithm in this paper => https://www.cs.purdue.edu/homes/hosking/690M/Bacon01Concurrent.pdf
12:03:44FromDiscord<flywind> It is slower than tracing GC like mark & sweep, ... when there are many cycles in your program.
12:08:50FromDiscord<sOkam!> I was considering them because they are recommended for latest code written in nim, and arc is specifically recommended for dynamic libraries (which is what I'm making, and it shares memory/data with C)
12:09:33FromDiscord<flywind> In reply to @sOkam! "Are all those calls": refc supports destructors too, though strings and seqs are not using destrctors.
12:09:38FromDiscord<sOkam!> But the documentation on the topic is all really advanced, and I'm having a lot of trouble learning how to use them in a practical day to day scrub-programmer type of way
12:11:38FromDiscord<sOkam!> Like, the intro for the destructors is literally a piece of code, that when you write it on a program it, the LSP immediately complains of duplication↵Does that mean that all code in that page is just illustrative of how the background code works... or does that mean that those implementations are the way you are supposed to do everything in your usual code?
12:13:28FromDiscord<sOkam!> - Do you need to use hooks manually for standard types?↵- Do you need to define hooks if you are inheriting from standard types?↵- Do you need to -use- hooks explicitely everywhere now that you are using that mm system... or is that supposed to be automatically done for you on compile?
12:13:52FromDiscord<sOkam!> None of those questions are really explained, so kinda lost on whether I should even use them... or how 😔
12:15:31FromDiscord<sOkam!> Easy choice is using refc, and GC_ref / unref.... but then... is that how you are supposed to interact with C code? That question is still unanswered for me, which is what made me start reading about orc/arc
12:15:49FromDiscord<Rika> its one way to do so, the ref unref
12:15:50FromDiscord<Rika> it works
12:16:18FromDiscord<sOkam!> whats the benefit on arc or orc then? and why are they recommended?
12:19:40FromDiscord<planetis> nope you can just ignore them
12:20:25FromDiscord<planetis> for optimization you can annotate sink parameters, although it's not always the right choice and lent return type
12:21:11FromDiscord<sOkam!> so you just write code as usual, and ignore the hooks? is that what you mean by "ignore them"?
12:21:31FromDiscord<Rika> you can ignore arc and orc as of now
12:22:13FromDiscord<flywind> Like lots of stdlibs still work without rewriting them with destructors.
12:22:27FromDiscord<sOkam!> then why are they recommended for all new code, and arc specially for dynamic libraries? 🤔
12:22:33FromDiscord<sOkam!> That's what I don't follow
12:25:00FromDiscord<planetis> there is breakage when transitioning for refc to orc in libs that deal with pointers and they ask to develop for orc first.
12:25:53FromDiscord<planetis> (edit) "for" => "from"
12:34:07FromDiscord<4zv4l> Hiii
12:34:26FromDiscord<4zv4l> how can I change the current process directory ? I looked into the doc but didn't find something similar to `chdir()`
12:35:53FromDiscord<4zv4l> dang
12:36:03FromDiscord<4zv4l> `setCurrentDir()` is what I need xD
12:41:18FromDiscord<sOkam!> In reply to @4zv4l "`setCurrentDir()` is what I": there is also `withDir` for temporary changes during a specific block of code. can be handy
12:46:09FromDiscord<Superstart033> Nim good?
12:51:19FromDiscord<4zv4l> In reply to @sOkam! "there is also `withDir`": oh thanks !
12:51:23FromDiscord<4zv4l> In reply to @Superstart033 "Nim good?": Nim good good
12:51:30FromDiscord<Superstart033> Thank you
12:52:04FromDiscord<4zv4l> is there an easy way to handle signals in Nim ? (SINGINT to be precise)
12:52:33FromDiscord<4zv4l> with sigaction or something idk
12:53:31FromDiscord<Rika> It’s in the POSIX module
12:53:34FromDiscord<Rika> Probably
12:55:02FromDiscord<4zv4l> what's the easiest pattern to check if I'm running on windows ?
12:55:20FromDiscord<4zv4l> because I guess if I try using sigaction and compile for windows I'll get an error
12:55:29FromDiscord<Rika> There’s a define switch but that’s only compile time, so it doesn’t detect if it’s under wine
12:56:01FromDiscord<AmitSupremacy> https://github.com/mratsim/Arraymancer/blob/master/examples/ex03_simple_two_layers.nim gives `Error: attempting to call undeclared routine: 'optimizer'`
12:56:35FromDiscord<4zv4l> In reply to @Rika "There’s a define switch": so if I compile it for windows I won't get error, it will just skip that part ?
12:57:08FromDiscord<Rika> If you use "when defined(posix): ..." I think
12:59:26FromDiscord<AmitSupremacy> In reply to @AmitSupremacy "https://github.com/mratsim/Arraymancer/blob/master/": anyone of arraymancer if can tell why is it not working
13:01:29FromDiscord<deeuu> In reply to @AmitSupremacy "anyone of arraymancer if": try in the #science channel
13:04:52FromDiscord<4zv4l> can someone make a small example of `sigaction` with Nim ? I can't get how it works
13:13:38FromDiscord<vindaar> which version of arraymancer are you on at the moment?↵(@AmitSupremacy)
13:16:07FromDiscord<AmitSupremacy> In reply to @vindaar "which version of arraymancer": #science please see
13:16:44FromDiscord<auxym> In reply to @4zv4l "can someone make a": quick google search turned this up, does it help? https://gist.github.com/dom96/908782?permalink_comment_id=2906764#gistcomment-2906764
13:21:03FromDiscord<d4rckh> after making a request using the http client to a host that is down, i keep getting `An existing connection was forcibly closed by the remote host.` on every single request to that host, even if it is up
13:24:19FromDiscord<d4rckh> reinitializing the http client seems to fix this
13:24:25FromDiscord<d4rckh> but why is it happening?
13:24:37FromDiscord<auxym> In reply to @d4rckh "after making a request": using which http client? Are you sure it's a nim issue? Did you try with curl?
13:24:53FromDiscord<d4rckh> its definitely a nim issue
13:24:55FromDiscord<d4rckh> std/httpclient
13:28:11FromDiscord<auxym> unless you get lucky and someone has already had this exact issue, your best bet to get some help would be posting full reproduction instructions (with code, maybe a local server that someone could bring up and down, etc) on the forum, I think.
13:41:03FromDiscord<4zv4l> In reply to @auxym "quick google search turned": well I try to understand the code but since I'm new to Nim it's not super easy xD
13:50:54FromDiscord<4zv4l> so
13:51:01FromDiscord<4zv4l> it works but doesn't prevent my program from stopping
13:51:11FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Vt
13:51:26FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Vu
13:59:52*bitgestalt joined #nim
14:10:46FromDiscord<4zv4l> well I fed up that idea xD I'll let SIGINT stop my program T-T
14:15:35FromDiscord<dom96> In reply to @d4rckh "after making a request": sounds like a bug, httpclient tries to reuse connections, it must be failing in this instance to create a new one
14:23:20FromDiscord<4zv4l> it there something similar to RAII in Nim ?
14:26:39FromDiscord<bren> I think destroy hooks, but AFAIK, these only work with arc/orc memory managers; you should be using one of the two though
14:26:55FromDiscord<bren> https://nim-lang.org/docs/destructors.html#lifetimeminustracking-hooks-nimeqdestroy-hook
14:27:31FromDiscord<4zv4l> `nim c --gc:arc --expandArc:all -d:release --opt:size -d:strip rev.nim && upx --best rev`↵yeah I use this
14:27:42FromDiscord<4zv4l> but so with this, it automatically add the `destroy(var)` ?
14:27:52FromDiscord<4zv4l> I don't need to add any free or something ?
14:28:10FromDiscord<4zv4l> I was looking to make a smaller binary but it seems it doesn't help xD
14:30:46FromDiscord<4zv4l> sent a code paste, see https://paste.rs/F1S
14:31:10FromDiscord<bren> zig doesn't link with libc btw, that's why it's size is so low
14:31:16FromDiscord<bren> (edit) "it's" => "its"
14:31:27FromDiscord<4zv4l> yeah also
14:31:34FromDiscord<4zv4l> but I wanted more to compete with C for that test
14:31:35FromDiscord<4zv4l> xD
14:31:40FromDiscord<auxym> had you seen this? https://hookrace.net/blog/nim-binary-size/
14:31:44FromDiscord<4zv4l> I know Zig size is really hard to beat
14:32:25FromDiscord<4zv4l> In reply to @auxym "had you seen this?": I have seen a similar post↵I did all except using the musl libc
14:33:51FromDiscord<bren> In reply to @4zv4l "`nim c --gc:arc --expandArc:all": I don't see passing flags to GCC here. Maybe you're doing it in a config file?
14:34:33FromDiscord<4zv4l> I don't use config file↵I'm really new so xD
14:34:54FromDiscord<!!sharpcdf!!> sent a code paste, see https://play.nim-lang.org/#ix=45VF
14:37:07FromDiscord<bren> In reply to @4zv4l "I don't use config": you can add a config.nims file to your root directory, and have the flags you want there instead of writing it in the terminal. Anyway, in that case, it means there are some flags you can pass to gcc to optimize more for size
14:38:24*arkurious joined #nim
14:39:33FromDiscord<4zv4l> In reply to @bren "you can add a": can I do that with the command line ?
14:39:54FromDiscord<Rika> In reply to @sharpcdf "my nimble file, ": Wrong type, not a seq but should be a string
14:40:16FromDiscord<bren> In reply to @4zv4l "can I do that": yup. the blog linked above does it that way
14:40:27FromDiscord<!!sharpcdf!!> In reply to @Rika "Wrong type, not a": ah i guess you're right, my bad 😅
14:42:31FromDiscord<4zv4l> alright I could go to 33k
14:44:04FromDiscord<4zv4l> `nim c --gc:arc --expandArc:all -d:release -d:strip --opt:size --passC:-flto --passL:-flto rev.nim && upx --best rev`↵using this command I go to 33k
14:44:22FromDiscord<4zv4l> how can I see if it handle well the memory because I didn't add myself the `destroy`
14:45:02FromDiscord<auxym> run in Valgrind or something I guess?
14:46:07FromDiscord<auxym> you shouldn't need destroy though unless you're doing something weird (or using pointers, manual allocations, etc). A memory leak in ARC would be considered a nim bug.
14:46:14FromDiscord<flywind> In reply to @4zv4l "`nim c --gc:arc --expandArc:all": you can also try -d:useMalloc
14:47:40FromDiscord<flywind> arc stands for automatically reference count, no need to manage the memory manually if sticking to destructors and refs.
14:47:57FromDiscord<flywind> (edit) "count," => "counting,"
14:48:12FromDiscord<4zv4l> so arc doesn't require a gc ?
14:48:35FromDiscord<bren> it's a "gc" itself. but a pretty light one
14:49:03FromDiscord<4zv4l> I thought it would add the destroy in the right place in the code before compiling
14:49:19FromDiscord<flywind> In reply to @4zv4l "so arc doesn't require": If you dont create cycles in your programs.
14:49:38FromDiscord<auxym> In reply to @4zv4l "I thought it would": yes, that is what it does (sort of)
14:49:51FromDiscord<4zv4l> In reply to @flywind "you can also try": yup from 33k to 30k
14:50:34FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45VI
14:50:50FromDiscord<4zv4l> because I don't see what's taking the space
14:51:15FromDiscord<4zv4l> if it compiles to c the code is almost exactly the same as the one I made in C
14:51:33FromDiscord<4zv4l> except the function name↵the logic is the exact same
14:52:40FromDiscord<auxym> well you're using a bunch of std libs. In C are you implementing all that stuff from scratch? your scratch implementation might be smaller (at the cost of less generality, etc)
14:53:00FromDiscord<4zv4l> also I just realized it isn't compatible with my c client↵like maybe it's the way the data is sent↵with nc it works for both c server and nim server
14:53:08FromDiscord<4zv4l> In reply to @auxym "well you're using a": hmmm I don't think I did that
14:53:57FromDiscord<auxym> the fmt macro in nim might be generating relatively large code, etc
14:54:06FromDiscord<4zv4l> https://media.discordapp.net/attachments/371759389889003532/1002227455517544548/server.c
14:54:10FromDiscord<4zv4l> thats the c server
14:54:21FromDiscord<4zv4l> In reply to @auxym "the fmt macro in": is there another way btw ?
14:54:55FromDiscord<4zv4l> In reply to @4zv4l "": like really the nim and c code are the same
14:55:30FromDiscord<auxym> yeah, directly wrap the low-level C functions (printf and getopt and whatnot) using importC. Probably smaller code but then you get all the unsafe and inelegant behavior from C.
14:55:57FromDiscord<4zv4l> Nim compiles to C right ?
14:56:24FromDiscord<4zv4l> then it could use the zig compiler which seems really efficient to make really small binaries
14:56:37FromDiscord<bren> yes nim compiles to C. That doesn't mean it uses the same stds
14:56:44FromDiscord<auxym> yes, but that doesn't mean it just uses everything from the C std lib. For example for strformat, nim has it's own implementation that safer and more convenient, it doesn't call `printf`
14:56:48FromDiscord<4zv4l> because between `gcc or tcc` and `zig cc` zig is really the one that make the smallest
14:57:25FromDiscord<4zv4l> can I get the c file generated ? I mean without looking into `~/.cache` ?
14:57:53FromDiscord<auxym> yes, by going into cache. you can modify the cache dir with a CLI opt to nim also.
14:58:35FromDiscord<auxym> the generated code isn't meant to be human readable though
14:59:10*noxnivi joined #nim
14:59:20noxnivigood day
14:59:34FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45VM
14:59:44FromDiscord<4zv4l> or it needs all the other file to compile ?
14:59:49FromDiscord<auxym> yes
14:59:53FromDiscord<auxym> and yes
14:59:53FromDiscord<bren> good day nox
14:59:56FromDiscord<4zv4l> xD
15:00:49FromDiscord<4zv4l> well 30k is already great
15:00:56FromDiscord<4zv4l> I meant
15:00:57FromDiscord<auxym> nim isn't just a "thin syntax layer" over C. It's a whole different language, that happens to use C as a backend target (instead of assembly/machine code). If you're trying to get "exactly the same code as hand made C" you'll be fighting a pointless battle.
15:01:02FromDiscord<4zv4l> small but great like for small size
15:01:23noxnivihi there bren
15:01:35FromDiscord<4zv4l> oh I didn't want to fight xD↵just wanted to get a smaller binary as possible
15:01:51FromDiscord<auxym> I think you have it 🙂
15:02:01FromDiscord<bren> In reply to @4zv4l "because between `gcc or": if you're willing to write everything from scratch, you can do --passL:-nostdlib --os:standalone --gc:none which doesn't link to libc. Then strip. You'll reach sizes smaller than zigs most likely, but good luck implementing an http server from scratch xD
15:02:22FromDiscord<4zv4l> In reply to @bren "if you're willing to": well then with zig you can also use `freestanding` xD
15:02:22FromDiscord<auxym> Going smaller at this point probably means not using the nim stdlib, writing your own implementations from scratch or calling straight into the C stdlib with importx
15:02:32FromDiscord<auxym> (edit) "importx" => "importc"
15:02:45FromDiscord<4zv4l> anyway the purpose isn't to make the smallest program possible↵as far as it works it's great
15:02:55noxnivii'm trying to build tinywm [the nim version
15:03:02noxnivias well as minimal
15:03:19noxniviand get an error that says 'cannot open file x11/xlib'
15:03:22noxnivion Arch linux
15:03:31noxnivithing is that Xlib is installed
15:03:35FromDiscord<4zv4l> I got it smaller for the `echo "hello, world !"` than in C↵but I guess it's not that easy xD
15:04:27noxniviat least presumably installed
15:04:44FromDiscord<bren> xlib-dev is installed right?
15:05:05FromDiscord<bren> (not using linux so idk what it's exactly named)
15:05:08noxnivithere is libx11, no xlib-dev available, at least pacman does not report such thing
15:05:19FromDiscord<bren> libx11-dev then
15:05:50noxnivino such thing neither, no libx11-dev
15:05:50FromDiscord<Rika> Arch does not use “-dev” packages I believe
15:06:15noxniviRika, afaik, you are right
15:08:01noxniviArch does not 'split' packages
15:08:06noxniviso with libx11 should be
15:08:17FromDiscord<bren> In reply to @Rika "Arch does not use": aha, I see
15:09:30FromDiscord<bren> yup that's right. I suppose you could find where the headers are and add it to include directories
15:12:24noxniviallright, located the lib
15:16:28*bitgestalt quit (Quit: Leaving)
15:23:36*LuxuryMode joined #nim
15:50:39noxnivithanks bren
15:50:48noxnivifinally could build minimal
15:50:53noxnivino success yet with tinywm
15:51:07noxniviplenty to learn yet
15:51:59FromDiscord<Goat> Theres a window manager using nim out there. I'm sure you could learn a lot by looking at the code
15:52:53noxniviyes, Goat, there is Nimdow and Worm, both available in Arch
15:53:05noxnivibut to start I rather go with the smallest ones
15:53:08noxnivi;}
15:58:45FromDiscord<sOkam!> Is there a way to create a C compatible struct, that will be shared between C and Nim, and accessed in both, without having to `importc`?↵I got recommended to importc, but I want to get rid of the C code, not depend on it. And just realized that importing it just makes me depend on it instead
16:04:42FromDiscord<domosokrat> `exportc` \:)
16:07:27*noxnivi quit (Remote host closed the connection)
16:27:15FromDiscord<sOkam!> In reply to @domosokrat "`exportc` \:)": and how do you know that the profile is gonna be compatible with C?
16:28:02*noxnivi joined #nim
16:28:10noxnivihi there again
16:30:29*Guest50 joined #nim
16:30:50FromDiscord<auxym> In reply to @sOkam! "and how do you": that's just a "contract" of the nim codegen. If you define an object in nim, the codegen will generate a struct with the same fields, types and field ordering in the generated C code. Or did I misunderstand the question?
16:31:57FromDiscord<sOkam!> In reply to @auxym "that's just a "contract"": by contract do you mean that its mandatory for the language to make structs that are compatible with C based on objects?
16:32:20FromDiscord<sOkam!> for a function, I had to make the exact signature and add some properties, so figured that i needed something like that for objects too
16:32:59FromDiscord<auxym> I'm not aware of any doc guaranteeing that. But at this point any change to this behavior would break most of the nim-c FFI currently existing.
16:33:27*Guest50 quit (Client Quit)
16:34:24FromDiscord<sOkam!> so if I add exportc and dynlib, like I did for the function, the C code will be able to read it as long as it has the same fields?
16:34:37FromDiscord<sOkam!> what about types. am I forced to use ctypes for the fields?
16:35:52*pch quit (Quit: Leaving)
16:37:30FromDiscord<auxym> you don't "have" to, but using types such as "cint" etc will make it easier for you to reason about what the codegen'd types will be (ie you can be 100% sure that `cint` in nim will be generated as `int` in C)
16:37:31FromDiscord<Phil> sent a code paste, see https://play.nim-lang.org/#ix=45Wm
16:38:34FromDiscord<Phil> I went full screaming case before but I'm none too happy with that either
16:38:35FromDiscord<auxym> using eg `int8` in nim woud be generated as some `_NI8` macro in C, defined somewhere in the nimbase header...
16:38:48FromDiscord<Arathanis> In reply to @auxym "you don't "have" to,": on that note, does nim's `uint64` translate to `uint64_t` from `stdint.h`?
16:39:22FromDiscord<auxym> In reply to @Isofruit "I went full screaming": I like pure enums a lot of the time. personal preference I guess
16:39:59FromDiscord<Phil> In reply to @auxym "I like pure enums": pure enum = all lowercase?
16:40:11FromDiscord<Phil> And no prefix?
16:40:18FromDiscord<auxym> In reply to @Arathanis "on that note, does": the C code will have `_NU64` in it IIRC, which is a macro that is platform dependent and I would assume it expands to `uint64_t` on most plateforms
16:40:30FromDiscord<sOkam!> For me, it depends on the use-case. If I plan for the enum field to be used separately, I name it `TheField`. But if I plan on using it with pure, then i use `TheEnum.field`
16:40:37*noxnivi quit (Remote host closed the connection)
16:41:06FromDiscord<auxym> In reply to @Isofruit "pure enum = all": see here the `.pure.` pragma https://nim-by-example.github.io/types/enums/
16:41:09FromDiscord<Phil> I feel like I stumbled into something that I wasn't aware before
16:41:12FromDiscord<Phil> Thanks
16:42:25FromDiscord<Phil> Ahh, so annotate enum with pure pragma and then PascalCase?
16:44:08FromDiscord<auxym> I usually use all lower, or snakecase, on the field names if I'm using pure. With the enum name itself having PascalCase
16:44:23FromDiscord<auxym> like `Color.red` or `Color.brightRed`
17:11:18FromDiscord<4zv4l> is there another way to format text (adding var into string)↵that including strformat↵and add `fmt""` ?
17:11:38*rockcavera joined #nim
17:11:39*rockcavera quit (Changing host)
17:11:39*rockcavera joined #nim
17:12:17FromDiscord<Rika> sure, `strutils` and `%`
17:12:27FromDiscord<4zv4l> which one is the most lighter ?
17:13:21*noxnivi joined #nim
17:18:56FromDiscord<Rika> light in terms of what
17:19:07FromDiscord<Rika> and i dont know, you should benchmark it yourself
17:19:14FromDiscord<Rika> it almost always depends per machine anyway
17:21:12FromDiscord<4zv4l> I'll stick with strformat for now I think↵easier to use and doesn't change the binary size
17:26:36FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Wv
17:28:53*noxnivi quit (Remote host closed the connection)
18:03:48NimEventerNew thread by Cnerd: Let variable is not gc safe, see https://forum.nim-lang.org/t/9331
18:13:35FromDiscord<voidwalker> How do you update an sql(ite) row that you got as the result of a SELECT query in instantRows or similar ? Do you have to put the unique key of the row in the SELECT statement and specify that in the UPDATE statement, or is there some shortcut ?
18:17:14FromDiscord<voidwalker> sent a code paste, see https://play.nim-lang.org/#ix=45WG
18:28:49*CyberTailor joined #nim
18:38:37FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45WR
18:39:16FromDiscord<4zv4l> do I need to precise the size or something ?
18:39:34FromDiscord<huantian> what is `Person`?
18:39:50FromDiscord<4zv4l> an object↵a struct
18:39:52FromDiscord<huantian> if it's a type and you want to use the type constructor, you have to do `Person(field: value)`
18:40:00FromDiscord<4zv4l> oooh
18:40:01FromDiscord<4zv4l> okok
18:40:22FromDiscord<4zv4l> can't it guess with the order ?
18:41:18FromDiscord<auxym> no, object construction in Nim is by named field only
18:41:44FromDiscord<huantian> you can make your own `proc initPerson` if you want
18:42:41FromDiscord<4zv4l> this still doesn't seem to work
18:42:41FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45WT
18:43:30FromDiscord<gibson> @4zv4l `Person(name: "Simon", age: 19, adr: "Rixensart")`
18:44:17FromDiscord<4zv4l> In reply to @gibson "<@329196212282458112> `Person(name: "Simon", age:": what if I want to make an array of struct ?
18:44:44FromDiscord<demotomohiro> let persons = [Person(name: "foo", age: 19), Person(name: "a")]
18:45:10FromDiscord<gibson> `@[Person(name: "Simon", ...), Person(name: "Ella", ...)` the "..." refers to like before. This is a list like a vector in c++. Use just `[Person(name: "Simon", ...), Person(name: "Ella", ...)` to make an array.
18:45:26FromDiscord<gibson> (edit) "...)`" => "...)]`"
18:45:32FromDiscord<gibson> (edit) "...)`" => "...)]`"
18:45:43FromDiscord<4zv4l> but with @[] it will be on the heap
18:45:49FromDiscord<4zv4l> but I already know the data and the size
18:46:07FromDiscord<4zv4l> so [person(), person(), ...] ?
18:46:09FromDiscord<gibson> So use the version that makes the array, without `@`.
18:46:12FromDiscord<gibson> Yep
18:46:29FromDiscord<4zv4l> sounds redundant but alright
18:48:40FromDiscord<gibson> sent a code paste, see https://play.nim-lang.org/#ix=45WW
18:50:40FromDiscord<gibson> @vzBut the "redundant" version lets you make a constant if that's desirable.
18:50:53FromDiscord<gibson> (edit) "@vzBut" => "@4zv4l But"
18:52:22FromDiscord<4zv4l> what do you mean ?
18:52:40om3gahi! what to do when compiler wont see cpp header?
18:53:46om3gait placed in subdir, and successfully compiled in linux
18:54:32FromDiscord<gibson> @4zv4l Const data is stored in a different location in RAM. Depends what you want to achieve. If you don't know the `const` keyword then don't worry about it and ignore me 🙂
18:54:37om3gabut in mac os it compiler throws err about unknown function: error: no matching function for call to 'r2c'
18:55:10om3gaI already tried to provide -I flag
18:55:55om3ganim bindings have {.pragma: pocket, header: pocketFFTPath / "pocketfft_hdronly.h".}
18:56:59FromDiscord<gibson> @om3ga At risk of being redundant, is `pocketFFTPath` pointing to the right place in mac os?
18:57:04FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45X3
18:57:31om3gagibson, yeah, it should. I tried to specify full path too
18:57:54FromDiscord<gibson> @om3ga you can print it out to verify.
18:58:54FromDiscord<gibson> @4zv4l Coming from python, maybe avoid arrays, and just use seqs for an introduction. They're much more flexible and behave much more like python lists.
18:59:04om3gagibson, pocketFFTPath = currentSourcePath.rsplit(DirSep, 1)[0] .. I will execute end print out this
18:59:25FromDiscord<4zv4l> I'm more coming from Zig and C and Ruby↵I actually dislike Python very much xD
18:59:28Amun-Ra4zv4l: array size must be known at compile time
18:59:49FromDiscord<gibson> @4zv4l I mean your friend is coming from python, so might be more at home with seqs.
18:59:52FromDiscord<4zv4l> yeah that's same in Zig but in C it's possible to have variable sized array on the stack
19:00:02Amun-Radepends on C standard
19:00:07FromDiscord<4zv4l> yeah true
19:00:11FromDiscord<4zv4l> but still possible
19:00:30Amun-Raseq in such VLA in Nim
19:00:39FromDiscord<4zv4l> VLA ?
19:00:44Amun-Ravariable length array
19:00:53FromDiscord<4zv4l> oh okok xD
19:01:09FromDiscord<gibson> You can have variable sized array on the stack, if you use the manual memory management, just like C. `alloc` `create` `dealloc`
19:01:18FromDiscord<4zv4l> isn't alloca possible also for dynamic array on the stack ?
19:01:38FromDiscord<gibson> well, heap, not stack anymore.
19:01:52Amun-Rathat's a pointer to be precise
19:01:53FromDiscord<4zv4l> alloca is to allocate more memory on the stack isn't it ?
19:02:36om3gagibson, yeah, seems fine
19:02:50om3gaany ideas guys?
19:03:14FromDiscord<4zv4l> anyway, could have been nice to be able to do like python for the struct array, so without having to name the field if they are in the right order
19:03:53FromDiscord<gibson> @4zv4l what's the python syntax you are trying to emulate? The nim version is basically adding `@` on the front.
19:04:14FromDiscord<4zv4l> for now that's what I have https://media.discordapp.net/attachments/371759389889003532/1002290402545631262/unknown.png
19:04:58FromDiscord<huantian> (btw usually we use `Natural` instead of `uint` in Nim)
19:05:13FromDiscord<huantian> you can emulate the python by making your own `initPerson` procedure
19:05:23FromDiscord<4zv4l> well Natural sounds weird and isn't common for people who doesn't know Nim so idk I use uint for now xD
19:05:29Amun-Rathat's possible for tuples
19:05:43FromDiscord<4zv4l> named tuple ?
19:06:00FromDiscord<huantian> sent a code paste, see https://play.nim-lang.org/#ix=45X6
19:06:27FromDiscord<4zv4l> thanks !
19:06:35FromDiscord<4zv4l> also if the proc doesn't return any value
19:06:40FromDiscord<4zv4l> do I need to precise `void`
19:06:48FromDiscord<4zv4l> or I can discard it ?
19:06:49Amun-Ra4zv4l: https://play.nim-lang.org/#ix=45X7
19:06:59FromDiscord<huantian> In reply to @4zv4l "do I need to": no
19:07:01FromDiscord<huantian> (edit) "https://play.nim-lang.org/#ix=45X6" => "https://play.nim-lang.org/#ix=45X8"
19:07:10FromDiscord<4zv4l> In reply to @Amun-Ra "<@329196212282458112>: https://play.nim-lang.org/#i": that's, actually great !
19:07:35FromDiscord<huantian> sent a code paste, see https://play.nim-lang.org/#ix=45X9
19:07:41FromDiscord<huantian> don't use tuples
19:07:58FromDiscord<4zv4l> In reply to @Amun-Ra "<@329196212282458112>: https://play.nim-lang.org/#i": wait but in the code here, does it use the Person tuple ?
19:08:03FromDiscord<4zv4l> since you don't precise it ?
19:08:16FromDiscord<huantian> all tuples are the same if it has the same types and names in the right order
19:08:24FromDiscord<huantian> which makes them bad for typing
19:08:28FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xb
19:08:47FromDiscord<4zv4l> In reply to @huantian "all tuples are the": so why precise the new type Person ?
19:08:59FromDiscord<huantian> `proc someProc(...) = ...` is the same as `proc someProc(...): void = ...`
19:09:07Amun-Ra4zv4l: and if you need object: https://play.nim-lang.org/#ix=45Xc
19:09:51FromDiscord<huantian> In reply to @4zv4l "so why precise the": because even if two objects have the same field types, you don't want to them to be treated the same
19:10:06FromDiscord<4zv4l> I mean
19:10:42FromDiscord<4zv4l> sent a code paste, see https://paste.rs/3va
19:10:43FromDiscord<huantian> yes
19:10:49Amun-Ra4zv4l: https://play.nim-lang.org/#ix=45X7
19:10:51FromDiscord<huantian> Person is just an alias for that tuple
19:10:51FromDiscord<4zv4l> so why making the type before ?
19:11:03FromDiscord<huantian> that way you don't have to type the type of the tuple each time you use it
19:11:06FromDiscord<huantian> you just use person
19:11:17FromDiscord<4zv4l> aaaaalright !
19:11:20FromDiscord<huantian> but again
19:11:24FromDiscord<huantian> don't use tuples
19:11:27FromDiscord<4zv4l> oh
19:11:29FromDiscord<4zv4l> why ?
19:11:37FromDiscord<4zv4l> In reply to @huantian "because even if two": here ?
19:11:40FromDiscord<huantian> yes
19:11:42Amun-RaI'd say use tuples if you don't need to modify them
19:11:51Amun-Rause objects otherwise
19:11:54FromDiscord<huantian> I'd say only use tuples as return types
19:11:56FromDiscord<huantian> and for unpacking
19:12:52om3gaah, I see it's Mac Os compiler issue. Please ignore, not nim related
19:13:11FromDiscord<4zv4l> also I didn't get the func/proc thing because↵I made a small code and couldn't use func even if the function was only printing to screen but it was making "side effects"
19:13:23FromDiscord<huantian> printing to the screen is a side effect
19:13:55FromDiscord<4zv4l> so better use proc everytime no ?
19:14:11Amun-Raif you need to "debug" something with echo inside a func use debugEcho
19:14:23FromDiscord<gibson> sent a code paste, see https://play.nim-lang.org/#ix=45Xd
19:14:24FromDiscord<huantian> sent a code paste, see https://paste.rs/Imm
19:15:01Amun-Ragibson: new is usually for refs
19:16:24FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xe
19:16:32FromDiscord<4zv4l> In reply to @gibson "<@329196212282458112> (discord crashed) you": I'll use something like that
19:17:35FromDiscord<4zv4l> thanks everyone anyway ! really friendly community
19:19:37FromDiscord<gibson> @4zv4l Amun-Ra is right, call it `init` instead of `new`.
19:19:39FromDiscord<Sun「無用」> Is there a way of mapping an array?
19:20:19FromDiscord<huantian> by map do you mean the function?
19:20:31FromDiscord<huantian> `sequtils.map` supports both arrays and seqs
19:21:00FromDiscord<Sun「無用」> In reply to @huantian "`sequtils.map` supports both arrays": and what is a seq or that sequtils
19:21:20FromDiscord<huantian> sequtils is part of the standard library
19:21:26FromDiscord<huantian> https://nim-lang.org/docs/sequtils.html
19:21:36FromDiscord<huantian> a `seq` is an array that can expand and shrink at runtime
19:21:41FromDiscord<huantian> arrays have to be fixed size
19:21:45FromDiscord<huantian> (edit) "arrays" => "`array`s"
19:22:13FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xg
19:22:22FromDiscord<4zv4l> with `nc` it works
19:22:27FromDiscord<4zv4l> but with the client it doesn't work
19:23:08FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xh
19:23:18FromDiscord<Sun「無用」> In reply to @huantian "https://nim-lang.org/docs/sequtils.html": and docs... I'm probably not understand it, but hx.
19:23:21FromDiscord<Sun「無用」> (edit) "hx." => "thx."
19:24:02FromDiscord<aMOPel> sent a code paste, see https://play.nim-lang.org/#ix=45Xi
19:24:19FromDiscord<huantian> out is a reserved keyword
19:24:27FromDiscord<huantian> it's not used for anything yet
19:24:38FromDiscord<aMOPel> alright 👍
19:25:20FromDiscord<aMOPel> it already is part of the syntax and compiles, though
19:26:05FromDiscord<Sun「無用」> how to turn a seq into a tuple/array?
19:26:21FromDiscord<4zv4l> sent a code paste, see https://paste.rs/thH
19:27:58FromDiscord<Sun「無用」> In reply to @Sun「無用」 "how to turn a": cause I didn't understand that seq docs, but it returns a seq and I have to handle it, unfortunately
19:46:35FromDiscord<Phil> In reply to @Sun「無用」 "how to turn a": First things first, do you know the length of that seq at compiletime?
19:51:22FromDiscord<Sun「無用」> In reply to @Isofruit "First things first, do": 2
19:51:30FromDiscord<Sun「無用」> probably
19:51:34FromDiscord<lantos> is there a nim package/framework that lets you define an API server and have the types/structures and requests available to be consumed by a client in TS/JS or used in something like OpenAPI?
19:51:50FromDiscord<lantos> (edit) "OpenAPI?" => "OpenAPI?↵↵Do the @status team have anything internally?"
19:52:31FromDiscord<locria> In reply to @Sun「無用」 "probably": You need like comptime and compileError
19:53:00FromDiscord<locria> code?
19:53:21FromDiscord<Sun「無用」> sent a code paste, see https://play.nim-lang.org/#ix=45Xq
19:53:43FromDiscord<4zv4l> any idea ?
19:53:44FromDiscord<4zv4l> sent a code paste, see https://paste.rs/dh0
19:53:59FromDiscord<4zv4l> it seems like the server doesn't know when to stop reading
19:54:11FromDiscord<locria> sent a code paste, see https://play.nim-lang.org/#ix=45Xr
19:54:12FromDiscord<locria> that's at runtime
19:54:32FromDiscord<Sun「無用」> In reply to @locria "what's the context": i've a seq, I want an arr or tuple.
19:54:38*pro joined #nim
19:55:00FromDiscord<Sun「無用」> used map on a array, or probably an array, dunno.
19:55:11FromDiscord<locria> In reply to @4zv4l "it seems like the": TCP?
19:55:54FromDiscord<locria> In reply to @Sun「無用」 "i've a seq, I": If the seq is constant, than make it array
19:56:07FromDiscord<4zv4l> In reply to @locria "TCP?": yeah
19:56:21FromDiscord<4zv4l> I also tried `recvline` but same result
19:56:38FromDiscord<locria> what is the code?
19:56:53FromDiscord<Sun「無用」> In reply to @locria "If the seq is": and I dunno how, lol.
19:57:03FromDiscord<locria> (edit) "what is the ... code?" added "complete"
19:57:09FromDiscord<4zv4l> the server
19:57:11FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xs
19:57:29FromDiscord<4zv4l> sent a code paste, see https://play.nim-lang.org/#ix=45Xt
19:57:54FromDiscord<4zv4l> it works with nc but except that with both the c client and nim client it doesn't work
19:58:12FromDiscord<locria> sent a code paste, see https://play.nim-lang.org/#ix=45Xu
19:59:40FromDiscord<Sun「無用」> and I didn't understand anything...
20:00:39FromDiscord<locria> check out slicing (search for `[0..` in Nim Manual)
20:01:05FromDiscord<4zv4l> In reply to @locria "what is the complete": does it help ? T-T
20:02:38FromDiscord<Sun「無用」> In reply to @locria "check out slicing (search": ik slicing, but okay
20:02:44FromDiscord<locria> In reply to @4zv4l "does it help ?": so
20:02:55FromDiscord<locria> your client doesn't send newline
20:03:06FromDiscord<locria> so the server never reads a line
20:03:34FromDiscord<locria> `send(server, input + "\n")`
20:03:47FromDiscord<Sun「無用」> In reply to @Sun「無用」 "ik slicing, but okay": I'm just dumb
20:04:12FromDiscord<locria> (edit) "+" => "&"
20:04:52FromDiscord<4zv4l> In reply to @locria "`send(server, input & "\n")`": still with this
20:04:55FromDiscord<4zv4l> it doesn't work T-T
20:06:10FromDiscord<locria> works for me
20:06:32FromDiscord<4zv4l> https://media.discordapp.net/attachments/371759389889003532/1002306079549554800/unknown.png
20:06:42FromDiscord<4zv4l> oooh
20:06:48FromDiscord<4zv4l> maybe it's the second readline from the client
20:06:49FromDiscord<4zv4l> I'll test
20:07:20FromDiscord<locria> I tried typing "bcri/,pntaosnhao" in client
20:09:32FromDiscord<treeform> You making your own remote shell? I find that using std input streams kind of difficult, they can be in terminal mode, buffer strangly etc...
20:09:50FromDiscord<treeform> When I made a shell I used a different protocol
20:09:56FromDiscord<treeform> And everything worked
20:10:14FromDiscord<locria> ~~or use tmux~~
20:10:36FromDiscord<locria> hello treeform. how's pixie v2?
20:11:00FromDiscord<treeform> Newline terminated protocols kind of suck.
20:11:14FromDiscord<treeform> In reply to @locria "hello treeform. how's pixie": You mean pixie 5?
20:11:18FromDiscord<4zv4l> In reply to @treeform "You making your own": I'm just trying to translate a program I had to make in C xD
20:11:19FromDiscord<locria> yeah
20:11:30FromDiscord<locria> In reply to @treeform "Newline terminated protocols kind": cough HTTP
20:12:03FromDiscord<treeform> New version of http is not newline terminated
20:12:21FromDiscord<locria> and based on UDP/QUIC
20:12:57FromDiscord<locria> TCP kind of suck in general
20:13:20FromDiscord<treeform> Yep
20:14:01FromDiscord<treeform> Pixie 5 has huge performance improvements making it faster or on par with cairo
20:14:16FromDiscord<treeform> But we are also removing masks
20:17:46FromDiscord<locria> still software rasterizer?
20:17:51FromDiscord<locria> (edit) "software" => "software-only"
20:19:53FromDiscord<4zv4l> I tried `output = recv(server, 4096)`↵but still doesn't work
20:20:03FromDiscord<4zv4l> it doesn't get the whole data sent by the servr
20:20:05FromDiscord<4zv4l> (edit) "servr" => "server"
20:20:49FromDiscord<4zv4l> or I make a thread that reads forever xD↵then I'm sure
20:22:29FromDiscord<dom96> In reply to @treeform "Newline terminated protocols kind": how so?
20:24:04FromDiscord<jan0809> i found them fun to play with tbh
20:24:23FromDiscord<jan0809> and faily ez to deal with
20:24:44FromDiscord<jan0809> (edit) "faily" => "fairly"
20:27:37*Guest42 joined #nim
20:27:44Guest42hi
20:27:48*Guest42 quit (Client Quit)
20:28:13FromDiscord<Phil> cheers
20:35:57FromDiscord<jan0809> hi
20:37:41FromDiscord<eyecon> sent a code paste, see https://play.nim-lang.org/#ix=45XH
20:39:12FromDiscord<eyecon> (edit) "https://play.nim-lang.org/#ix=45XH" => "https://play.nim-lang.org/#ix=45XI"
20:39:20FromDiscord<4zv4l> In reply to @4zv4l "I tried `output =": it works in C so why not Nim ? T-T so tired of it
20:58:33FromDiscord<jan0809> sent a code paste, see https://play.nim-lang.org/#ix=45XN
21:02:03*jmdaemon joined #nim
21:05:04FromDiscord<4zv4l> Do like my c program xD
21:05:09FromDiscord<4zv4l> Read as much and print what it got
21:05:15FromDiscord<4zv4l> No need to wait to fill the buffer
21:06:07FromDiscord<4zv4l> In reply to @4zv4l "that's how I read": This works with the c server
21:06:12FromDiscord<4zv4l> Even if it doesn't fill the buffer
21:09:47*pro quit (Quit: pro)
23:11:57FromDiscord<!Patitotective> how do i measure the "runs" per second in a while loop?
23:13:54FromDiscord<!Patitotective> more like iterations per second
23:16:11FromDiscord<Hi02Hi> well runs/second is 1/(seconds/run), and seconds per run is just timing how long it takes, you can use something like benchy
23:22:42FromDiscord<!Patitotective> sent a code paste, see https://play.nim-lang.org/#ix=45Yi
23:22:44FromDiscord<!Patitotective> https://github.com/Patitotective/downit
23:27:27FromDiscord<!Patitotective> sent a code paste, see https://play.nim-lang.org/#ix=45Yj
23:29:03FromDiscord<Prestige> could the server be throttling you? Probably better to test with a local server to prevent any weirdness
23:31:48FromDiscord<!Patitotective> In reply to @Avahe "could the server be": i dont think it is↵its always the same, no matter which url i use
23:32:33FromDiscord<!Patitotective> In reply to @Avahe "could the server be": anyway how should i test it with a local server? 🤨
23:32:57FromDiscord<Prestige> You'd write a server to run on your computer and send a request to it
23:33:34FromDiscord<!Patitotective> ive no idea how to write a server :[
23:34:12FromDiscord<Prestige> https://nim-lang.org/docs/asynchttpserver.html
23:34:29FromDiscord<Prestige> Maybe doesn't matter if the url changing doesn't matter for your example
23:34:46FromDiscord<Prestige> but I'd just do it as a practice if you're trying to test downloads
23:37:55FromDiscord<!Patitotective> In reply to @Avahe "https://nim-lang.org/docs/asynchttpserver.html": amazing
23:41:30FromDiscord<!Patitotective> ok so, using the code from the example it returns `Hello World` when `curl localhost:1234`
23:41:45FromDiscord<!Patitotective> how would i use it as a url to download something form it?
23:41:47FromDiscord<!Patitotective> (edit) "form" => "from"
23:46:38FromDiscord<!Patitotective> using `waitFor sleepAsync(1)` works perfectly