<< 21-08-2018 >>
00:02:09vivuswho uses choosenim here?
00:14:38FromDiscord<emekoi> i use choosenim
00:16:34*xet7 quit (Quit: Leaving)
00:33:13FromGitter<kaushalmodi> @xmonader I'm going to go through it and understand the "graph theory" better. It's like a 'make' clone, right?
00:33:46FromGitter<kaushalmodi> Also, in the comments you mention possible future support of YAML. Have you considered TOML? :)
00:33:57vivuson linux? @emekoi
00:34:51FromGitter<kayabaNerve> vivus I do on Linux
00:35:08FromGitter<kaushalmodi> @PMunch and @data-man have worked on updating the Nim TOML parser (nimble package).
00:35:39vivus@kayabaNerve is the installation as simple as it sounds? just run the bash script?
00:35:57FromGitter<kayabaNerve> Yep
00:37:22FromGitter<xmonader> @kaushalmodi Yes, it's more like a very minimalistic version of how make works ⏎ ⏎ and for the yaml sugestion ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5b7b5ec138a12915e4d237a7]
00:37:39FromGitter<xmonader> it's just for describing the the tasks without worrying about special language
00:37:44FromGitter<xmonader> toml would work too
00:41:36FromGitter<retsyo> what a bad local internet connection I have. I can onyl download nim 0.17.2 < 10KB/s
00:43:21*vivus quit (Quit: Leaving)
01:11:19*Jesin quit (Quit: Leaving)
01:14:44*Jesin joined #nim
01:50:28*vivus joined #nim
01:53:15FromGitter<retsyo> nim 0.17.2 can compile wxnim's example without being stuck, although there are too many errors so that no exe can be produced. Nim Compiler Version 0.18.1 (I compiled from github's code at 2018-08-18) always be stuck. There is still a long time before I can transport my wxPython application to nim
01:59:27*cspar quit (Ping timeout: 240 seconds)
02:07:21*ftsf joined #nim
02:43:50*cspar joined #nim
02:53:41shashlick_@retsyo: is it Araq's version or PMunch's fork
02:53:44shashlick_https://github.com/PMunch/wxnim
03:01:55*dddddd quit (Remote host closed the connection)
03:38:46*hogeland joined #nim
03:40:01FromGitter<retsyo> PMunch's fork
03:45:11*endragor joined #nim
04:12:51*Vladar joined #nim
04:26:55*vivus quit (Remote host closed the connection)
04:32:38*ftsf_ joined #nim
04:34:50*ftsf quit (Ping timeout: 244 seconds)
05:25:05*nsf joined #nim
05:55:19*ftsf joined #nim
05:58:32*ftsf_ quit (Ping timeout: 244 seconds)
06:03:38*stefanos82 joined #nim
06:07:43*Trustable joined #nim
06:56:51FromGitter<timotheecour> how can i check if a symbol has a pragma? eg: `inline` (I see `hasPragma` in compiler/pragmas.nim, but that’s not available in libraries eg standard library)
06:57:17*dadabidet joined #nim
07:03:30Araqyou can't.
07:03:53FromGitter<timotheecour> fundamental limitation or fixable?
07:05:29Araqwe can add/patch macros.hasPragma but I'm not sure I want somebody's DSL to inspect my .inline annotations
07:08:40FromGitter<timotheecour> eg use case: I need a trait isInline so I can customize code depending on whether iterator is `inline` or `closure`
07:09:47FromGitter<xmonader> guys isere ssh library for nim or do i need to implement it :( ?
07:15:08*PMunch joined #nim
07:17:25Araqok, sounds reasonable
07:17:56Araqxmonader: not sure
07:21:49FromGitter<xmonader> I see a great potential for nim if we have ssh library, we can basically overthrow ansible and all automation frameworks outthere with speed/macro systems what do you guys think?
07:24:43AraqI wrapped https://github.com/Araq/bearssl
07:25:02Araqbut haven't used it for anything, be my guest and adopt it
07:25:47FromGitter<xmonader> @Araq this is great thansk a lot
07:27:02*Trustable quit (Remote host closed the connection)
07:35:20*norok2 quit (Quit: Gone hunting.)
07:38:14livcdxmonader: there's a wrapper for ssh afaik
07:47:28*norok2 joined #nim
08:03:38*gmpreussner joined #nim
08:04:21FromGitter<bung87> how to access object property using string key?
08:04:54*gmpreussner_ quit (Ping timeout: 272 seconds)
08:05:32FromGitter<timotheecour> use `fieldPairs`
08:06:26*Jesin quit (Ping timeout: 276 seconds)
08:08:03FromGitter<bung87> https://nim-lang.org/docs/system.html#fieldPairs.i,T this?
08:08:11FromGitter<timotheecour> yes
08:08:26FromGitter<bung87> ok,thanks ! let me try it
08:11:18FromGitter<bung87> https://github.com/nim-lang/Nim/pull/6987 I see this PR which in devel branch,it's more easy
08:14:22FromGitter<timotheecour> how is that related to `to access object property using string key` ? `fieldPairs` will be more general (will work with any existing object / tuple)
08:16:48FromGitter<bung87> the serializationKey does not for this?
08:17:57*Jesin joined #nim
08:18:22FromGitter<timotheecour> i don’t think so, i think that’s a custom thing; feel free to try with `type A=object (a:int)`
08:22:22FromGitter<bung87> proc `[]`(x: BMES,index:string): float = ⏎ result = MIN_FLOAT ⏎ for name, value in x.fieldPairs: ⏎ ⏎ ```when name == index: ⏎ return value``` ... [https://gitter.im/nim-lang/Nim?at=5b7bcbbd38a12915e4d497c3]
08:23:02FromGitter<bung87> cannot evaluate at compile time: index ⏎ ,what's wrong with the code?
08:23:47FromGitter<timotheecour> when expects CT input, index is RT
08:24:15leorizeuse if instead
08:24:30FromGitter<alehander42> and you dont need return result
08:24:40FromGitter<alehander42> result is returned by default
08:25:49FromGitter<timotheecour> if you’re trying to have polymorphic output (eg return int or float depending on field), this won’t work, for that you’d need basically what I implemented (which is like D’s `static foreach`) ; haven’t sent a PR for it yet
08:26:54FromGitter<bung87> hmm ,yesterday I write some code it's doest return by default,I know it will return as last time I wrote nim,I can't confirm now
08:27:47FromGitter<bung87> oh I 'll check what's the diffriences between CT and RT
08:27:59FromGitter<timotheecour> compile time vs run time
08:28:45FromGitter<alehander42> if you have a result variable and you haven't returned until the end of function, result will be returned
08:29:30FromGitter<bung87> thanks guys,learn a lot.
08:30:11FromGitter<timotheecour> Pleasure :)
08:30:26FromGitter<bung87> I'm porting a python package to nim,but it results 7 times slower than python,I may face more problems...
08:31:39FromGitter<timotheecour> if you can reduce your test case, feel free to post it in forum or even (if you’re quite sure it’s not normal) in github issues
08:34:53FromGitter<bung87> https://github.com/bung87/finalseg/blob/master/src/finalseg.nim ,yeah I pushed it to github in minutes ago,I'm now trying cast JsonNode to object
08:35:31FromGitter<tim-st> use `packedjson` instead of `json`
08:37:13FromGitter<mratsim> don’t declare var result and then return result
08:37:29FromGitter<mratsim> result is implicit, already available and returned automatically
08:37:29FromGitter<tim-st> also be careful with rune procs, e.g. `runeLen`is O(n) operation
08:38:14FromGitter<tim-st> instead often `fastRuneAt` works
08:38:39*krux02 joined #nim
08:39:11FromGitter<bung87> ok,I 'll check these things
08:39:18FromGitter<tim-st> did you compile in release mode?
08:39:21FromGitter<mratsim> you should us getOrDefault here: https://github.com/bung87/finalseg/blob/master/src/finalseg.nim#L47-L48
08:39:25*Vladar quit (Remote host closed the connection)
08:39:44FromGitter<mratsim> same: https://github.com/bung87/finalseg/blob/master/src/finalseg.nim#L61 getOrDefault
08:41:16FromGitter<bung87> yes release mode,but didnot reduce much time than debug mode,in release mode it's about 7 seconds,but python results about 2 seconds
08:41:44FromGitter<mratsim> states and PrevStatus should be an enum: https://github.com/bung87/finalseg/blob/master/src/finalseg.nim#L43
08:42:18FromGitter<mratsim> you allocate too much strings, Python GC will reuse those, but Nim just return and re-request memory from the system
08:42:39FromGitter<tim-st> ` if sentence.runeLen() == 0:` is only true if `sentence.len == 0` is true, too while last one is O(1)
08:42:40FromGitter<mratsim> unlike Python, Nim GC will not cache memory that can be reused
08:43:26FromGitter<mratsim> don’t allocate memory in tight loops: https://github.com/bung87/finalseg/blob/master/src/finalseg.nim#L53-L56, clear the table and reuse.
08:44:00FromGitter<mratsim> already said but states should be enum, not strings: https://github.com/bung87/finalseg/blob/master/src/finalseg.nim#L98
08:44:01FromGitter<tim-st> also if you want "Han" use `\p{Han}` I think
08:44:17FromGitter<bung87> wow so many,let me check it one by one....
08:45:41FromGitter<alehander42> also you do a lot of JsonNode getFloat in tight loops, I suspect it might be way faster to just parse the json files once into Nim objects/tables and directly use them
08:46:01FromGitter<bung87> I dont have confidence about `\p{Han}` whether contains tradition chinese or not,I 'll dig it later
08:46:38FromGitter<mratsim> Basically, there are 2 things you should be aware of to get the fastest speed: ⏎ ⏎ 1) use O(1) algorithms like tim-st said ⏎ 2) avoid allocating memory in loops: strings, tables, json, seq all allocates. Appending or updating an existing one is fine, but creating a new one is very costly. [https://gitter.im/nim-lang/Nim?at=5b7bd16e94f8164c177acfdf]
08:47:33FromGitter<tim-st> also: when you create new seq or new string, often you know nearly the capacity, though the python code doesnt do it too so it's not the bottleneck
08:47:45FromGitter<bung87> @alehander42 yes,you are right,I also have another branch which not complete,since it's not like dict in python.
08:48:11FromGitter<tim-st> so you can do `result = newStringOfCap(input.len div 3)` or similiar
08:50:05FromGitter<mratsim> now the last thing is that you probably need to add `{.noInit.}` to each proc that returns a string or a seq similar to this: https://github.com/mratsim/Arraymancer/blob/master/src/tensor/init_cpu.nim#L49. ⏎ ⏎ Unfortunately Nim is a bit too conservative at the moment with seq/strings results and does a costly extra initialization pass even when you initialize the results in the proc body :/.
08:50:14FromGitter<bung87> `newStringOfCap` where to use it?
08:50:48FromGitter<tim-st> also if you care about peformance and you have such easy regex: iterate over runes and compare the rune Range manual using `rune.ord in 0xXXX .. 0xYYYY`
08:51:00FromGitter<tim-st> e.g. `var result = newSeqstring ()`
08:51:11FromGitter<tim-st> this has no capacity and uses cap = 8
08:51:43FromGitter<tim-st> replacing regex by manual runes checking will make your program like 10 times faster than python
08:52:29*nedbat23 joined #nim
08:54:02*mappum joined #nim
08:55:56FromGitter<bung87> yeah ,there's also a cpp version of this package,It does use rune iterator,I dont know that parts have so much performance cares,thanks told me this.
08:57:18*mappum quit (Remote host closed the connection)
08:57:18*nedbat23 quit (Read error: Connection reset by peer)
09:07:13*fredrik92 is now known as couven92
09:08:53*skelett joined #nim
09:09:22FromGitter<bung87> @mratsim if I use enum how do i using string key?
09:15:24FromGitter<tim-st> you use the chars of strings as states, so you can create `type State = enum of A,B,B` and then for state in State: let c = state.ord.char; but this wont change performance
09:15:36FromGitter<tim-st> 1) `A,B,C`
09:16:16FromGitter<mratsim> @bung87 why do you need string keys?
09:17:50FromGitter<mratsim> string are very slow. It’s a pointer (8-bytes) + length (8 bytes) + storing the string data (variable size).
09:18:01FromGitter<mratsim> an enum is just 1 byte and always on the stack
09:18:34FromGitter<mratsim> strings are on the heap meaning Nim has to ask the OS “Please give me some memory”, this is VERY slow.
09:18:39FromGitter<tim-st> I think he uses const string in the example
09:19:15FromGitter<mratsim> the left hand side is allocated at runtime
09:19:29FromGitter<mratsim> and even with const string there is pointer dereferencing to read the data.
09:20:00FromGitter<tim-st> ok
09:20:42FromGitter<tim-st> but I think the biggest performance impact will have handling runes correctly, replacing regex by rune checking, and replacing `json` by `packedjson`
09:21:58FromGitter<mratsim> I’m not specialist of that, but I’m pretty sure I could improve perf by 10x~20x with much better allocations and ading {.noInit.}
09:22:04FromGitter<bung87> I checked the getOrDefault it's return JsonNode,it has to cast to float.
09:22:54FromGitter<tim-st> On my lmdb nim can iterate over more than 300k k, v allocated strings by copy so allocating strings is maybe bad regarding performance, but wont have the impact in the example code compared to python I think
09:23:14FromGitter<tim-st> (300k per second)
09:23:37FromGitter<mratsim> for arraymancer, allocating seq is the biggest perf bottleneck
09:24:11FromGitter<tim-st> yes, I can imagine, but your code is already very optimized ;-)
09:24:22FromGitter<mratsim> especially when converting a tensor to string, that’s so slow :/
09:24:34FromGitter<tim-st> once regex is involved, there are other bottlenecks
09:24:41FromGitter<mratsim> yeah I can imagine
09:25:52FromGitter<tim-st> btw I tried to cast a `UncheckedArray[char]` to string where I know startadress and len of chars, should this work (casting to native nim string)?
09:26:57FromGitter<tim-st> even in Python they can do zero copy from lmdb :\
09:27:26FromGitter<bung87> I'm not familar with program language that have to care about allocating,I will check json and regex at first
09:27:51FromGitter<mratsim> @tim-st no it won't
09:28:01FromGitter<tim-st> why not?
09:28:25FromGitter<tim-st> I saw an example of casted seq in `system.nim` I copied the coded but read from nil it said
09:28:34FromGitter<mratsim> the memory is like this \[ length (8 bytes) | start of memory blob (length bytes)\]
09:28:49FromGitter<mratsim> you will be missing the length part
09:29:30FromGitter<tim-st> I copied this https://github.com/nim-lang/Nim/blob/b75808c7d992ea47f8d6abc656b881b2aa0f86de/lib/system.nim#L696
09:29:47FromGitter<tim-st> The type `PGenericSeq` should do it, I thought
09:29:58FromGitter<mratsim> result is a seq
09:30:13FromGitter<mratsim> and you cannot just cast unmanaged memory to GC memory
09:30:28FromGitter<bung87> hmm I know that but it's still difficult for me to manually manage allocating,I usauly write js,ts and python codes.
09:30:37FromGitter<tim-st> @mratsim https://github.com/nim-lang/Nim/blob/b75808c7d992ea47f8d6abc656b881b2aa0f86de/lib/system.nim#L426 has the string definition
09:30:58FromGitter<tim-st> I thought about it as unsafe string; the memory is freed by lmdb after transaction is commited
09:31:07FromGitter<tim-st> this is how python does it too
09:31:27FromGitter<mratsim> @bung87 you don’t need to manually manage allocating memory, this is another discussion. But you should avoid creating strings, seq and tables in a tight loop.
09:33:15FromGitter<mratsim> @tim-st so I’m wrong, it’s \[length (8 bytes) | reserved memory (8 bytes) | data (length bytes) \]
09:34:00FromGitter<tim-st> So should it be possible to cast to the definition I linked, under the assumption that everything is freed by external library?
09:34:03Araqmratsim: please give me a snippet where .noInit is required
09:34:42AraqI don't want Nim programmers to write .noInit
09:34:43FromGitter<mratsim> @Araq, here you go: https://github.com/frol/completely-unscientific-benchmarks/pull/17
09:35:15Araqthat's a PR, not a snippet
09:35:47FromGitter<mratsim> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5b7bdcf3fd3c641b06eae1c9]
09:36:27FromGitter<mratsim> 2x speed improvement for `time main` when using noInit
09:37:18FromGitter<mratsim> only for the refcounting GC, the mark and sweep doesn’t have the same issue
09:37:32Araqthat makes no sense, looking into it
09:38:05FromGitter<mratsim> it’s related to this fix: https://github.com/nim-lang/Nim/commit/88cf6573e04bd7ee8762aa336460b9748f0d4644
09:38:33PMunchAnyone know what's the root of the issue here? https://stackoverflow.com/questions/51932881/calling-fzf-from-nim-doesnt-show-fzfs-tui/51938887#51938887
09:38:44FromGitter<mratsim> resetLoc shouldn’t be used for seq/string/ref that are properly initialized within the proc body.
09:38:58FromGitter<mratsim> i.e. when the compiler can prove that they are not nil?
09:41:18Araqthe only noInit that you need is here
09:41:19Araqproc splitBinary(orig: Node, value: int32): (Node, Node) {.noinit.}
09:45:40FromGitter<mratsim> right, confirmed on my machine. I have to check why I need to sprinkle it everywhere in arraymancer, I never return nil.
09:46:35Araqit's indeed genericReset
09:46:53Araqwhich is avoided for pure 'ref's but not much else...
09:46:58FromGitter<mratsim> or maybe it’s containers
09:47:08FromGitter<mratsim> I see, well seq are impacted badly
09:56:38*TheLemonMan joined #nim
10:04:25endragorha - "[GC] cannot register global variable; too many global variables", and the limit is hardcoded...
10:04:41FromGitter<ZarsBranchkin> How big is the limit?
10:05:22endragor3499 (interesting number btw). It's quite reachable, if you use `unittest` module in a large project.
10:06:41endragor3500 in fact
10:07:51FromGitter<ZarsBranchkin> How exactly you hit it? All tests in one file or what
10:12:21*Vladar joined #nim
10:12:29Araqendragor: unittest should introduce helper procs but you can also do that
10:12:33PMunchWhy is there a limit?
10:13:23endragorZarsBranchkin: there is a main file which imports modules with test suites. The limit is not per-module, but per-project.
10:13:38Araqthe GC cannot use 'seq' internally as seqs are GC'ed
10:14:15endragorAraq: could it use manually-managed dynamic array?
10:14:46Araqsure but the limit is fine, we need to fix unittest
10:15:12FromGitter<ZarsBranchkin> I don't see the limit bad as well, you don't really want to have packed global scope anyway
10:15:40Araqproc helper() {.gensym.} = testBody
10:15:41FromGitter<ZarsBranchkin> But that is annoying that unittest runs into such problem
10:15:44Araqhelper()
10:15:52Araqsomewhere in unittest.nim would do the job
10:31:12*SenasOzys__ quit (Ping timeout: 272 seconds)
10:31:50*SenasOzys quit (Ping timeout: 272 seconds)
10:31:51*SenasOzys__ joined #nim
10:35:58*SenasOzys joined #nim
10:42:13*Tux26 joined #nim
10:42:45FromGitter<gogolxdong> How to express a or b notin c?
10:45:14*Tux26 quit (Remote host closed the connection)
10:46:13FromGitter<gogolxdong> a notin c and b notin c?
10:49:10Araqc * {a, b} == {}
10:52:14FromGitter<gogolxdong> if a,b,c is string
11:21:53TheLemonMandata-man, I've just added a brief mode that's IMO nicer than tabulated output
11:24:20*Vladar quit (Remote host closed the connection)
11:24:44*Vladar joined #nim
11:31:04FromGitter<data-man> @TheLemonMan: Nice! But I dream about more formatters: markdown, rst, html, ... :)
11:35:49TheLemonMandata-man, the idea is to export everything to csv and have a karax+d3 based app to show pretty graphs :)
11:38:19FromGitter<data-man> Oh! I like libs without other depedencies :(
11:41:31TheLemonManI'm more a "do one thing and do it well" guy heh
11:43:25*dadabidet quit (Quit: Leaving)
11:44:19stefanos82TheLemonMan: the UNIX philosophy, hm?
11:46:40FromGitter<mratsim> @TheLemonMan, I’m thinking on creating a Vega library on top of Arraymancer, check Vega it seems great: https://vega.github.io/, and this video at 19:27 - https://youtu.be/GdoDLuPe-Wg?t=1167
11:55:42*SenasOzys quit (Ping timeout: 260 seconds)
11:56:12*SenasOzys__ quit (Ping timeout: 268 seconds)
12:04:01*dddddd joined #nim
12:07:23*SenasOzys joined #nim
12:07:36*SenasOzys__ joined #nim
12:29:18*xet7 joined #nim
12:32:25FromGitter<mratsim> was there like +5 backers in the past month? https://github.com/nim-lang/Nim#backers
12:32:45*seni joined #nim
12:35:06TheLemonManmratsim, that's nice, thanks for the link
12:36:23FromGitter<mratsim> polestar would be the equivalent to Tableau
12:36:39FromGitter<mratsim> (Tableau was codenamed Polaris in the early days apparently)
12:37:47TheLemonMana ggplot-inspired library written in Nim would also be nice to have
12:38:33dom96https://nim-lang.org/sponsors.html
12:43:25euantorI stopped sponsoring because I didn't want to support Bounty SOurce anymore and never knew there was an OpenColelctive. Guess I'll sign back up now :)
12:46:47yglukhov[m]Araq: what do you think about the `--experimental:forLoopMacros` problem? are you aware of it? cc @mratsim
12:47:14FromGitter<mratsim> @yglukhov, we discussed it in the Status slack ;)
12:47:56yglukhov[m]o rly. let me find it...
12:48:43FromGitter<mratsim> we will have the same issue if we depend on `—experimental:notnil`, report here: https://github.com/nim-lang/Nim/issues/8691 and https://github.com/nim-lang/Nim/issues/8676
12:49:34FromGitter<mratsim> putting the flag in a nim.cfg contaminates building `koch` for csources during CI.
12:49:59FromGitter<mratsim> alternative is to always use `nimble test` and put the switch in the corresponding task
12:50:25FromGitter<mratsim> it makes developing a bit more annoying because you have to remember to have your own nim.cfg with the flag on
12:50:27yglukhov[m]imo, `nim.cfg` doesn't even have to be considered as a solution to this problem. merely a workaround.
12:51:08FromGitter<mratsim> the main issue is that {.experimental:Foo.} doesn’t work properly for macro with generics input
12:52:05FromGitter<mratsim> well, it’s not that it doesn’t work, but that it’s required at all places that call the macro or a generic proc that uses the macro
12:52:42*arlen joined #nim
12:52:59FromGitter<mratsim> see here and the next 2 replies: https://github.com/nim-lang/Nim/issues/8676#issuecomment-413989546
12:53:41PMunchHmm, given a series of consecutive IDs ranging from 1 to highestID. How can I create all unordered sets containing up to N items without repetition?
12:53:46PMunchAnyone got any clever ideas?
12:54:32FromGitter<mratsim> nCk?
12:55:14FromGitter<mratsim> https://en.wikipedia.org/wiki/Binomial_coefficient
12:56:00FromGitter<mratsim> I think Rust has a macro called cartesian_product that almost does what you need (it’s not up to N it consumes all values)
12:56:03PMunchAh yes
12:56:35FromGitter<mratsim> Python also has itertools.product it seems
12:56:50PMunchBasically I want to iterate over all the possibilities, so I don't need to actually create a list of these
12:57:27*arlen quit (Remote host closed the connection)
13:00:24PMunchPython's product doesn't seem to be unordered
13:00:50PMunchie. product('ABCD', repeat=2) would produce both AB and BA
13:01:41PMunchIt also has combinations though, which seems to be what I want
13:01:52PMunchAB AC AD BC BD CD
13:03:31PMunchTo generate up to N I could just run it with increasing numbers
13:05:52*noonien joined #nim
13:10:16*cryptocat1094 joined #nim
13:12:48FromGitter<mratsim> then look into the difference between cartesian combinations and permutations
13:13:02FromGitter<mratsim> nCk vs nPk or nCr vs nPr I don’t remember
13:13:23PMunchI think I found something I can use
13:13:33PMunchhttps://web.archive.org/web/20120213125655/http://www.merriampark.com/comb.htm
13:17:42*lucy_ joined #nim
13:19:16FromGitter<data-man> @PMunch ⏎ Something like this? ⏎ ⏎ ```import algorithm ⏎ echo product([@['A', 'B'], @['A', 'B']])``` [https://gitter.im/nim-lang/Nim?at=5b7c115338a12915e4d63c42]
13:21:17PMunchNot quite
13:21:25PMunchI need it without duplicates and without ordering
13:21:37PMunchThe input AB only has one possibility AB
13:21:47PMunchABC has AB, AC, and BC
13:22:00PMunchABCD has AB AC AD BC BD CD
13:22:26*lucy_ quit (Remote host closed the connection)
13:27:47*endragor quit (Remote host closed the connection)
13:29:24FromGitter<data-man> https://bitbucket.org/nimcontrib/nimcombinatorics/src
13:29:35FromGitter<data-man> @PMunch ^
13:31:08FromGitter<ephja> will non-ordinal type discriminators never be supported? I'm trying to implement zero-cost abstractions for interfacing with C again
13:32:21FromGitter<mratsim> you mean object variant?
13:33:16PMunchdata-man: ah that's what I was looking for
13:33:57PMunchOr well, optimally it would return me a set
13:34:07PMunchBut I can work with this :)
13:36:19FromGitter<ephja> @mratsim yes, but I think that acceptable workarounds exist
13:38:02FromGitter<mratsim> No I meant, what does a non-ordinal type discriminator look like, it it like an object variant (with the discriminator being say an enum with holes or that start with negative value)
13:39:09TheLemonMandata-man, #7867 is fixed, rejoice! :D
13:42:01*nsf quit (Quit: WeeChat 2.2)
13:43:20FromGitter<data-man> @TheLemonMan I love you! :-D
13:44:54FromGitter<ephja> @mratsim oh. yes, object variants with discriminators containing holes. The usual C business
13:45:37*endragor joined #nim
13:49:34FromGitter<mratsim> I’m pretty sure there was a regression open on that
13:49:40*PrimHelios joined #nim
13:50:18*endragor quit (Ping timeout: 260 seconds)
13:53:09FromGitter<mratsim> sorry I’m confusing with sets, see https://github.com/nim-lang/Nim/issues/8340, you can use a macro to fill holes: https://github.com/status-im/nimbus/blob/master/nimbus/vm/interpreter/utils/macros_gen_opcodes.nim#L25 @ephja
13:54:19FromGitter<mratsim> and see discussion here: https://github.com/nim-lang/Nim/pull/8264
13:55:33FromGitter<ephja> right. that complicates the interface though and isn't feasible for large holes
14:04:02FromGitter<ephja> having separate type classes for distinguishing between enums and discriminators with and without holes at compile time seems like a good solution to this problem, but manpower remains a scarce resource ;)
14:07:40Araqenums with holes should be mapped to 'distinct int' or similar
14:08:06AraqNim has enough features to model everything in a pretty precise way
14:09:21*Hijiri28 joined #nim
14:10:42*Hijiri28 quit (Remote host closed the connection)
14:14:33FromGitter<ephja> that is an improvement over plain integer constants, but a worse improvement than is theoretically possible :-P
14:15:58FromGitter<xmonader> Guys, are there any notes on day 11? something vague, or difficult? something can be better? I think the path calculation can be better using a stack instead of parentMap table ..
14:16:52FromGitter<ephja> but I'm going to have a go at some improvements involving metaprogramming (yes, really!). I think there are decent solutions
14:21:06*PMunch quit (Quit: Leaving)
14:30:29FromDiscord<awr> i've been thinking of trying to make bindings for apple's WebGPU
14:30:34FromDiscord<awr> for nim
14:31:34*salewski joined #nim
14:32:42FromGitter<ephja> Yep, a flag type still seems like a good idea. sure, you can have sets with large values (usually because of enums with holes), but the memory usage is going to explode
14:33:24FromGitter<mratsim> @awr, is it even used, the only project I know that uses it is a deep learning project: https://github.com/mil-tokyo/webdnn but only being able to run on Safari is a bit meh
14:34:26*salewski quit (Client Quit)
14:52:45FromGitter<Clyybber> Araq what makes https://github.com/nim-lang/Nim/issues/7816 so tough to fix?
14:53:09FromGitter<Clyybber> Or where does the problem stem from?
14:54:50*ftsf quit (Ping timeout: 244 seconds)
14:54:55AraqI don't remember, sorry :-)
14:55:15FromGitter<Clyybber> ok no problem
15:02:39FromDiscord<awr> @mratsim WebGPU is a draft proposal and is the only one with any (publicly available) implementation
15:02:46FromDiscord<awr> there's a few other ones
15:04:52FromDiscord<awr> IMO the Obsidian one looks better https://github.com/KhronosGroup/WebGLNext-Proposals/tree/master/Obsidian-Mozilla as it seems WebGPU is missing a few things (there doesn't appear to be a device object apparatus in WebGPU like VkDevice or ID3D12Device and if I'm not mistaken resembles the OpenGL way of going about things)
15:05:25FromDiscord<awr> but AFAIK there is no implementation of it
15:06:12FromDiscord<awr> https://github.com/KhronosGroup/WebGLNext-Proposals/tree/master/Obsidian-Mozilla#screen-clear this resembles Vulkan a lot
15:07:05FromDiscord<awr> plus i don't think WebGPU uses SPIR-V which is 😐
15:07:40FromGitter<mratsim> I’m fine with Vulkan-like stuff, Apple has a very annoying tendency of not supporting standards it pushed in the long run: OpenCL, OpenGL, Webkit
15:08:26Araqcan we please stop turning the browser into an OS. ugh. wasm, webGPU
15:08:28FromGitter<mratsim> I also remember the fight to get hardware accelerated h264 working on Mac on 3rd party applications ...
15:08:56FromGitter<mratsim> Tell that to Chromebooks and FirefoxOS ;)
15:09:30Araqthese are all far worse than even Posix is.
15:09:32FromGitter<mratsim> oh, and Electron ugh
15:09:40FromDiscord<awr> @Araq have you seen this https://www.destroyallsoftware.com/talks/the-birth-and-death-of-javascript
15:09:55Araqawr: yes. it's pretty old.
15:12:49livcdlooks like an irreversible process
15:19:18FromGitter<ephja> run!
15:29:06*endragor joined #nim
15:32:28*endragor quit (Remote host closed the connection)
15:41:58*tiorock joined #nim
15:42:00*tiorock quit (Changing host)
15:42:00*tiorock joined #nim
15:42:00*rockcavera quit (Killed (orwell.freenode.net (Nickname regained by services)))
15:42:00*tiorock is now known as rockcavera
15:46:21*CompanionCube26 joined #nim
15:46:32*shadowbane quit (Read error: Connection reset by peer)
15:48:28*CompanionCube26 quit (Remote host closed the connection)
15:48:52*shadowbane joined #nim
15:57:17*shadowbane quit (Quit: Konversation terminated!)
15:57:36*shadowbane joined #nim
15:58:14*shadowbane quit (Client Quit)
15:58:33*shadowbane joined #nim
16:00:34FromGitter<arnetheduck> in nim csources, there's both a `build.sh` and a `makefile` - why both (on unix) / why is `build.sh` preferred? technically, a makefile could for example be used to parallelize initial build
16:03:42*cspar quit (Ping timeout: 272 seconds)
16:14:17dom96A long time ago the Nim c sources contained only a build.sh file
16:14:31dom96Then somebody came along and said "I ran make and it failed, I'm so used to that, you should have a makefile!"
16:14:45dom96So they created a PR and it was merged
16:14:55FromGitter<kaushalmodi> I think that just a Makefile should suffice
16:15:03dom96Since then the presence of both files has confused many
16:16:31FromGitter<kaushalmodi> Btw seems like it's mostly me talking about Nim on Mastodon. Folks there should use the #nim tag for more visibility
16:19:29dom96I guess I should set up a nim-lang mastodon account and somehow mirror Twitter there
16:20:26FromGitter<kaushalmodi> dom96: Yes
16:21:08FromGitter<kaushalmodi> mastodon.technology instance has a nice tech-focused crowd
16:38:53*SenasOzys quit (Ping timeout: 260 seconds)
16:39:28*SenasOzys__ quit (Ping timeout: 260 seconds)
16:41:12*druonysus joined #nim
16:41:13*druonysus quit (Changing host)
16:41:13*druonysus joined #nim
16:43:40TheLemonMandom96, are you able to reproduce #8211?
16:45:45dom96Nope, looks like it's fixed
16:45:56dom96Wanna make a PR to add it as a test case?
16:59:05shashlick_Use tissue
17:01:25shashlick_Super easy to make an issue snippet into a test case
17:03:35*ng0 quit (Remote host closed the connection)
17:04:11*SenasOzys joined #nim
17:04:24*SenasOzys__ joined #nim
17:06:12*Taylor_ joined #nim
17:09:46*PrimHelios quit (Ping timeout: 260 seconds)
17:10:22*zachk joined #nim
17:11:34*zachk quit (Changing host)
17:11:34*zachk joined #nim
17:19:04*ng0 joined #nim
17:29:20*sz0 joined #nim
17:31:36*NimBot joined #nim
17:40:55*stefanos82 quit (Quit: Quitting for now...)
17:44:14*nsf joined #nim
18:06:08*vivus joined #nim
18:21:31*PMunch joined #nim
18:23:47PMunchUhm why am I getting "Error: unhandled exception: n < 21 43 is too large to look up in the table [AssertionError]" when running fac?
18:23:57PMunchIt appears to calculate things in math.nim
18:26:57FromGitter<mratsim> fac is recalculated up to 20
18:27:03FromGitter<mratsim> precalculated
18:27:08FromGitter<mratsim> more will overflow
18:28:47FromGitter<mratsim> factorial 43 is 60415263063373835637355132068513997507264512000000000
18:28:55PMunchAh, that makes sense..
18:29:02FromGitter<mratsim> ~6.10^52
18:30:58PMunchI was trying to calculate factorial(43/(10*factorial(42 - 10))
18:31:06PMunchNo wait
18:31:33PMunchfactorial(43)/(factorial(10)*factorial(42 - 10))
18:32:17FromGitter<mratsim> https://github.com/numforge/number-theory/blob/master/src/combinatorics.nim#L4-L7
18:33:08FromGitter<mratsim> binomialCoeff(43, 10) should give you the result
18:34:11*Jesin quit (Quit: Leaving)
18:34:32Araqyeah, don't use the primitive definition to compute binom()
18:35:01PMunchI copied that code from somewhere else
18:35:21PMunchAnd thanks for the procedure mratsim, it works wonders :)
18:36:44PMunchI think I might have to find a better way to compute this though :P
18:37:32*Connor[m]1 quit (*.net *.split)
18:37:32*endes[m] quit (*.net *.split)
18:37:32*Miguelngel[m] quit (*.net *.split)
18:37:32*opi quit (*.net *.split)
18:37:32*zahary_ quit (*.net *.split)
18:37:33*dashed quit (*.net *.split)
18:37:33*dvn quit (*.net *.split)
18:37:33*Elronnd quit (*.net *.split)
18:37:33*vqrs quit (*.net *.split)
18:37:33FromGitter<mratsim> what do you want to compute in the end?
18:37:41*zahary_ joined #nim
18:37:48*opi- joined #nim
18:37:56*dashed joined #nim
18:38:20PMunchThe optimal list of N ingredients to make the most IBA cocktails
18:38:35*Miguelngel[m] joined #nim
18:38:55PMunchThis is what I've got so far: http://ix.io/1kSw
18:39:15*Connor[m]1 joined #nim
18:39:16*endes[m] joined #nim
18:39:21PMunchdrinks.toml: http://ix.io/1kSx
18:39:34TheLemonManisn't this a knapsack problem?
18:39:42PMunchingredients.toml: http://ix.io/1kSy
18:39:57*elronnd joined #nim
18:40:16PMunchHmm, kinda
18:40:36PMunchBut how do you weight the items?
18:40:48*elronnd is now known as Elronnd
18:41:17federico3it's not knapsack
18:42:02TheLemonManI don't know what IBA is but I assumed it has to do with the most alchool you can cram in a glass :)
18:42:16*TheManiac[m] quit (*.net *.split)
18:42:16*xomachine[m] quit (*.net *.split)
18:42:16*toofly[m] quit (*.net *.split)
18:42:18federico3TheLemonMan: it's not
18:42:39FromGitter<dlutton> definitely not knapsack
18:42:40PMunchhttps://en.wikipedia.org/wiki/List_of_IBA_official_cocktails
18:42:49PMunchIt's the International Bartenders Association
18:43:12TheLemonManoh, I see
18:43:41*xomachine[m] joined #nim
18:43:42*TheManiac[m] joined #nim
18:43:54*toofly[m] joined #nim
18:44:27*d10n quit (Ping timeout: 240 seconds)
18:46:08*dvn joined #nim
18:47:08*vqrs joined #nim
18:47:15*d10n joined #nim
18:47:25*d10n quit (Changing host)
18:47:25*d10n joined #nim
18:47:55TheLemonManso you want to compute the minimal set of ingredients that includes the most recipes (sets) as possible?
18:48:33*rockcavera quit (Remote host closed the connection)
18:48:47FromGitter<mratsim> @PMunch do you know if there is an optimal solution? Otherwise i would bruteforce my way through: https://github.com/Vindaar/nimnlopt
18:49:48FromGitter<mratsim> L-BFGS is a good algorithm to start with
18:49:50PMunchTheLemonMan, yes basically. I'm planning to add in things like ingredients you have, filtering out things like the "Miscellaneous" ingredients I have in my list, etc.
18:50:10PMunchmratsim, well what do you mean optimal?
18:50:19PMunchI guess there would be multiple equally good solutions..
18:50:46FromGitter<mratsim> Or the go player in me would suggest Monte-Carlo Tree Search
18:51:58PMunchHmm, so far it's 1% through the list of "Unforgettables" with 10 ingredients..
18:52:05FromGitter<mratsim> optimal like, it’s a solved problem or can be considered solved (like the travelling salesman problem)
18:52:27FromGitter<mratsim> https://en.wikipedia.org/wiki/Travelling_salesman_problem
18:52:50PMunchNot sure TBH
18:52:58PMunchI haven't done a whole lot of research into the topic..
18:53:40FromGitter<mratsim> I find your problem similar to N-Queens
18:53:45TheLemonManyou may gain some speed by parallelizing the checks
18:54:09PMunchTheLemonMan, oh yeah, for sure
18:54:55TheLemonManbut the search space is quite big, if you don't really need _the_ optimal solution you may try a random approach
18:55:44TheLemonManmaybe a genetic algorithm whose fitness function is based on the number of drinks that need a given ingredient
18:57:57PMunchHmm, that could work..
18:57:59federico3PMunch: I'm pretty sure there are already lists like this for IBAs. Anyhow if you want a perfect solution you have to iterate over all possible N-out-of-M combinations
18:58:35PMunchfederico3, oh yeah that's probably true. But really besides the point
18:58:41PMunchI want to have a general solution
18:58:50PMunchN-out-of-M is basically what I'm doing now
18:59:23PMunchI think with filtering random ingredients and running multi-threaded it won't be as bad
18:59:28federico3otherwise you could sort the M possible ingredients based on how many cocktails you can make with them, then pick the first K, with N < K < M and do combinations only on those
19:01:24PMunchYeah I thought about something like that
19:04:36FromGitter<mratsim> You should look into Monte-Carlo Tree Search. It will always give you the best solution given the time budget you give it, and it converges to the optimal solution.
19:04:36*Jesin joined #nim
19:04:59PMunchHmm, that is interesting
19:05:46FromGitter<mratsim> Your problem looks very similar to Multi-Armed Bandit https://en.wikipedia.org/wiki/Multi-armed_bandit ⏎ ⏎ > In probability theory, the multi-armed bandit problem (sometimes called the K-[1] or N-armed bandit problem[2]) is a problem in which a fixed limited set of resources must be allocated between competing (alternative) choices in a way that maximizes their expected gain, when each choice's properties are only
19:05:46FromGitter... partially known at the time of allocation, and may become better understood as time passes or by allocating resources to the choice.
19:06:45FromGitter<mratsim> (Go is a multi-armed bandit problem and Monte-Carlo Tree Search brought lots of improvement to it)
19:07:40PMunchHmm
19:08:27FromGitter<ephja> yay convergence!
19:08:38FromGitter<mratsim> if you’re looking for implementations there are several for tic-tac-toe in Python that you can just lift.
19:09:20TheLemonManthere's also this nice site/book on the topic http://banditalgs.com/
19:13:03FromGitter<mratsim> @TheLemonMan, nice! The exercises sound fun
19:13:27*elrood joined #nim
19:17:59TheLemonManmratsim, I sent a patch for #8706, can you give it a spin when you have some time?
19:19:48TheLemonManthe silly empty node was also bothering me to no end
19:20:31Araqany ideas what to do with .experimental?
19:21:09AraqI can only think of bad internal compiler complexity
19:21:18*Jesin quit (Quit: Leaving)
19:21:48FromGitter<mratsim> what happened?
19:23:02Araqmratsim: well you want .experimental to work across generic instantiations
19:23:57FromGitter<timotheecour> @araq are you talking about https://github.com/nim-lang/Nim/issues/8676 ?
19:24:20Araqyes
19:24:41FromGitter<mratsim> for now we cope, nimbus and stint are all green. But we have to use nimble switches. Nim.cfg leaking during CI is annoying.
19:25:00Araqoh yglukhov already found a nice solution. hmm
19:25:19Araqbut that would only work for forLoopMacros
19:26:21FromGitter<timotheecour> so u also want `caseStmtMacros` right?
19:26:31FromGitter<mratsim> experimental macros are tricky with generics and it’s fine if there is a solution that only affect macros
19:26:44FromGitter<mratsim> now what about experimental: “NotNil”?
19:29:44*Vladar quit (Remote host closed the connection)
19:32:25*Jesin joined #nim
19:33:40Araqwhat about it?
19:33:59*xomachine[m] quit (*.net *.split)
19:33:59*Connor[m]1 quit (*.net *.split)
19:33:59*zielmicha[m] quit (*.net *.split)
19:34:00*subsetpark quit (*.net *.split)
19:34:01*JStoker quit (*.net *.split)
19:34:10zachkwhat does the var do in this procedure type signature? : proc events(pressed: var seq[sdl.Keycode]): bool =
19:34:34*xomachine[m] joined #nim
19:34:39FromGitter<timotheecour> u can modify `pressed`
19:34:52*subsetpark joined #nim
19:34:54*zielmicha[m] joined #nim
19:36:35*JStoker joined #nim
19:37:02*Connor[m]1 joined #nim
19:38:14FromGitter<mratsim> @Araq, are there corner cases for experimental “NotNil"
19:38:41*macsek1911[m] quit (Ping timeout: 255 seconds)
19:39:11*federico3 quit (*.net *.split)
19:39:12*SunDwarf quit (*.net *.split)
19:39:12*ltr_ quit (*.net *.split)
19:39:12*Calinou quit (*.net *.split)
19:39:22zachkdoes it modify the variable passed into the procedure, so the changes to it are accessible once the procedure returns?
19:39:34*Calinou joined #nim
19:39:34*SunDwarf joined #nim
19:40:37*macsek1911[m] joined #nim
19:43:48FromGitter<Varriount> zachk: It allows the original location of the variable to be redefined.
19:44:22FromGitter<Varriount> If you are familier with C or C++, it's roughly equivalent of passing a pointer to the stack location of the variable.
19:45:24*rockcavera joined #nim
19:46:26*skelett1 joined #nim
19:46:38*federico3 joined #nim
19:48:23*sz0 quit (Quit: Connection closed for inactivity)
19:48:26*skelett quit (Ping timeout: 256 seconds)
19:48:27*cow_2001 quit (Ping timeout: 256 seconds)
19:49:37*ltr_ joined #nim
19:56:44*arecaceae quit (Remote host closed the connection)
19:57:03*arecaceae joined #nim
20:01:17*cow_2001 joined #nim
20:02:27FromGitter<ephja> @mratsim I dunno, do the bugs qualify? :p
20:11:12Araqmratsim: probably. the problem is that we don't know if we flag the declaration site or the usages with .experimental
20:11:26*jeff_ joined #nim
20:11:53*jeff_ is now known as Guest4113
20:14:37*livcd_ joined #nim
20:15:02*Rush joined #nim
20:17:10*shodan45_ joined #nim
20:18:23*qwertfisch_ joined #nim
20:19:27*dddddd quit (*.net *.split)
20:19:28*qwertfisch quit (*.net *.split)
20:19:28*shodan45 quit (*.net *.split)
20:19:28*gangstacat quit (*.net *.split)
20:19:29*RushPL quit (*.net *.split)
20:19:29*livcd quit (*.net *.split)
20:24:06*dddddd joined #nim
20:24:57*gangstacat joined #nim
20:45:08*Guest4113 quit (Quit: Page closed)
20:48:43*qwertfisch_ left #nim (#nim)
20:49:49*krux02 quit (Remote host closed the connection)
20:50:28cow_2001if you import everything into the global namespace, how do you know which name belongs to which module?
20:53:04shashlick_That's not required
20:53:15*cryptocat1094 quit (Quit: WeeChat 2.2)
20:53:23PMunchcow_2001, in the future? IDE tools
20:54:49FromGitter<mratsim> works already in VS code
20:55:11cow_2001can I import like they do in Python?
20:55:15FromGitter<mratsim> yes
20:55:18cow_2001cool!
20:55:23FromGitter<mratsim> from foo import nil
20:55:30FromGitter<mratsim> and you must qualify everything
20:55:38FromGitter<mratsim> but it breaks uniform call syntax
20:56:16FromGitter<mratsim> or just `import foo` and qualify when you want
20:59:57*smt joined #nim
21:01:02FromGitter<data-man> @Araq: https://github.com/nim-lang/Nim/pull/8718 Excuse me for delay, please!
21:01:36*qwertfisch joined #nim
21:08:07*qwertfisch quit (Quit: ZNC - http://znc.in)
21:08:48*qwertfisch joined #nim
21:12:52*data-man joined #nim
21:15:41*data-man quit (Client Quit)
21:15:58*data-man joined #nim
21:17:44*data-man quit (Client Quit)
21:18:16*data-man joined #nim
21:18:43*druonysus quit (Read error: Connection reset by peer)
21:19:06*seni quit (Quit: Leaving)
21:19:37*data-man quit (Client Quit)
21:19:59*nsf quit (Quit: WeeChat 2.2)
21:44:48*rockcavera quit (Remote host closed the connection)
21:48:29*PMunch quit (Remote host closed the connection)
21:49:23FromGitter<arnetheduck> @dom96, any preference between shell and makefile? I guess shell script could simply call make, for compatibility.. can't imagine a unix with shell and gcc but without makefile, so it shouldn't bring any incompatibilities
21:50:17dom96shell is a POSIX standard. Can the same be said for makefiles?
21:50:50dom96or at least, can we be sure that the makefile will be compatible with the same or more platforms as shell?
21:50:57copygirlI want to write an event system in a game, where any object could be fired as an event.
21:51:00dom96This includes exotic things like Haiku
21:51:49copygirlFor example, simply defining a `type MouseDownEvent* = object; position*: Point; button*: MouseButton`
21:53:05copygirlI would like to have two generic functions, ideally called as `MouseDownEvent.subscribe(HandleMouseDown)` and similar for `unsubscribe`.
21:54:40copygirlI was wondering if I could set things up in such a way that each concrete implementation of the generic function would come with its own sequence of handlers. Or well.. each pair.
21:55:13copygirlI rememebr the `Type.` thing being possible, though I don't remember how exactly.
21:55:44dom96proc subscribe(self: typedesc, prc: proc ())
21:56:13copygirlAnd I'm somewhat certain Nim is also able to do static members in a generic proc..?
21:56:30dom96But doing it this way is a little misleading because you're not saving the handler inside the type
21:56:43dom96You need an object to save these handlers into
21:57:48*elrood quit (Quit: Leaving)
21:57:57copygirlSomething like `proc subscribe[T](self: typedesc[T], prc: proc(T))` wouldn't work? How do I tell the handler it must accept an argument of type T?
22:03:07*SenasOzys quit (Read error: Connection reset by peer)
22:03:30*SenasOzys joined #nim
22:03:37*SenasOzys__ quit (Ping timeout: 268 seconds)
22:03:54*SenasOzys quit (Remote host closed the connection)
22:04:20*SenasOzys joined #nim
22:08:38*rockcavera joined #nim
22:16:30*xet7 quit (Quit: Leaving)
22:17:33FromGitter<arnetheduck> yeah, it's part of posix: http://pubs.opengroup.org/onlinepubs/9699919799/
22:22:53*TheLemonMan quit (Quit: "It's now safe to turn off your computer.")
22:38:12copygirlI would welcome any pointers regarding my questions. Please highlight me. I will be off for now. Thank you!
22:41:31copygirlhttps://github.com/nim-lang/Nim/issues/6406 <- Actually I did find this, so that's awesome! But some hints about creating a per generic implementation static field would be nice.
22:45:11*druonysus joined #nim
22:45:12*druonysus quit (Changing host)
22:45:12*druonysus joined #nim
22:57:00*and joined #nim
22:59:02*and quit (Remote host closed the connection)
23:11:12*casdr12 joined #nim
23:14:12*casdr12 quit (Remote host closed the connection)
23:20:17*cspar joined #nim
23:29:07FromGitter<ephja> is semantic highlighting implemented in any editor other than vim?
23:38:13*ftsf_ joined #nim
23:42:25FromGitter<rayman22201> semantic highlighting? do you mean syntax highlighting? Vim, Emacs, and VSCode all have working syntax highlighting for Nim.
23:45:25*zachk quit (Quit: Leaving)
23:54:38FromGitter<ephja> semantic syntax highlighting to be exact (per-variable highlighting, per-symbol type highlighting and so on)
23:55:16FromGitter<ephja> probably with the help of nimsuggest, unless some mad genius has created a viable alternative by now 😛