<< 10-06-2018 >>
00:00:43*thomasross joined #nim
00:32:20*SenasOzys quit (Ping timeout: 245 seconds)
00:57:15*krux02_ quit (Remote host closed the connection)
01:01:29*CodeVance quit (Read error: Connection reset by peer)
01:05:42*CodeVance joined #nim
01:05:43*CodeVance left #nim (#nim)
01:36:40*PMunch quit (Quit: leaving)
01:48:23*donlzx joined #nim
02:12:25*}{ultimo}{ left #nim (#nim)
02:12:51*lainon joined #nim
02:59:37*CodeVance joined #nim
03:03:21CodeVanceHow to kill a thread
03:03:45CodeVanceCan't find any info on it
03:08:32*CodeVance quit (Read error: Connection reset by peer)
03:09:21*CodeVance joined #nim
03:22:06*CodeVance left #nim (#nim)
03:35:18*yglukhov[i] joined #nim
03:39:27*yglukhov[i] quit (Ping timeout: 240 seconds)
03:52:58*CodeVance joined #nim
04:31:20*cspar quit (Ping timeout: 276 seconds)
04:56:12*cspar joined #nim
05:20:53*SenasOzys joined #nim
05:33:22*CodeVance quit (Quit: Leaving.)
05:35:32*miran joined #nim
05:38:30shashlickPMunch: this bug - https://github.com/nim-lang/Nim/issues/7776, cc @mratsim
06:44:52Araqyou can't "kill" a thread, the posix APIs pretty much deprecated it
06:45:06Araqyou thread needs to support receiving a "shutdown" message
06:45:14Araqor else you use a process instead.
06:48:22*xkapastel quit (Quit: Connection closed for inactivity)
06:55:30*jjido joined #nim
06:58:15FromGitter<Varriount> Anyone have a clue as to why https://github.com/nim-lang/Nim/pull/8002 has failing checks?
07:07:17*nsf joined #nim
07:16:48FromGitter<tim-st> Varriount: manual_snippet_104.nim and tut1_snippet_116.nim raise Errors
07:17:27FromGitter<tim-st> and also some more related to these files
07:26:38*donlzx quit (Remote host closed the connection)
07:44:25*jjido quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
07:51:32*subsetpark quit (Ping timeout: 260 seconds)
07:59:14*subsetpark joined #nim
08:04:17*gmpreussner quit (Ping timeout: 248 seconds)
08:05:15*gmpreussner_ joined #nim
08:11:07*tzekid joined #nim
08:28:29*gangstacat quit (Quit: Ĝis!)
08:36:27*yglukhov[i] joined #nim
08:39:22*xet7 joined #nim
08:42:54Araqtim-st: these are "correct" failures, should probably change 'nim doc' so that they don't appear in the logs
08:49:43*dddddd joined #nim
08:59:05*lainon quit (Ping timeout: 240 seconds)
09:00:18*BitPuffin joined #nim
09:14:48*Vladar joined #nim
09:17:20*yglukhov[i] quit (Remote host closed the connection)
09:18:53*gangstacat joined #nim
09:25:59*jjido joined #nim
09:26:28*skrylar quit (Remote host closed the connection)
09:29:04*yglukhov[i] joined #nim
09:32:47FromGitter<tim-st> ok, thanks!
09:33:57FromGitter<tim-st> btw are there cases when `nim check file.nim` finds errors that are not found by nimsuggest?
09:49:57*SenasOzys quit (Ping timeout: 240 seconds)
09:50:45*jjido quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
09:51:12*krux02 joined #nim
10:00:02*Shivelight joined #nim
10:01:55Araqtim-st: nimsuggest has the occasional glitch because it needs to be fast
10:02:17Araqso 'nim check' is more reliable. but incremental compilation should fix that
10:05:02FromGitter<tim-st> ok, thanks, good to know
10:07:33*SenasOzys joined #nim
10:09:58Araqbtw 'nim check --os:linux'/macosx/windows is underused
10:10:17Araqbut works really well for cross platform development
10:14:40*tzekid quit (Ping timeout: 256 seconds)
10:31:09FromGitter<Vindaar> @Araq in case you're still around: I've been looking at https://github.com/nim-lang/Nim/issues/7997 to learn a little more about the compiler. The issue of course is that `locks` defines a `release` proc. So that after symbol lookup the `release` in `defined(release)` is of type `nkSym` instead of `nkIdent`. Would a fix now try to make sure it's still of type `nkIdent` or simply handle `nkSym` in `semexprs.semDefined`?
10:32:14FromGitter<cschardt> cs @2vg: cool thing. Question: how did you create the page and particularly its navigation frame?
11:14:47*elrood joined #nim
11:25:47AraqVindaar: just handle nkSym in semDefined, excellent digging! :-)
11:34:39*yglukhov[i] quit (Remote host closed the connection)
12:11:26FromGitter<Vindaar> @Araq Ok, I did that. Running tests now. I'm not sure under which circumstances the `"obsolete usage of 'defined', use 'declared' instead"` was supposed to appear though. So maybe I'm shadowing some of these cases with my inclusion of nkSym now?
12:18:10*PMunch joined #nim
12:19:28Araqyou can actually remove this error message
12:19:37Araqdefined vs declared is a clear distinction now
12:20:12FromGitter<Vindaar> Ah, even better! Should I just discard all cases besides `nkIdent` and `nkSym` then?
12:20:26Araqfor 'defined', sure
12:20:39FromGitter<Vindaar> ok!
12:21:12Araqor better use lookups.considerQuotedIdent
12:22:19FromGitter<Vindaar> yeah, that's already in use for the check against `"defined"`: ⏎ ⏎ ```if not onlyCurrentScope and considerQuotedIdent(c.config, n[0], n).s == "defined":``` [https://gitter.im/nim-lang/Nim?at=5b1d17fbffd8896fe919c0a5]
12:23:24*find0x90 joined #nim
12:25:08*yglukhov[i] joined #nim
12:28:20FromGitter<Vindaar> here we go: https://github.com/nim-lang/Nim/pull/8005
12:41:44*sz0 quit (Quit: Connection closed for inactivity)
12:49:13dom96Varriount: Did you see my PM?
13:04:23dom96huh, when did this get released? https://narimiran.github.io/nim-basics/
13:04:29FromGitter<Varriount> Ah, just saw it.
13:04:54mirandom96: yesterday, but it is in "beta-phase"
13:05:04dom96looks good
13:05:11dom96Mind if I tweet about it? :)
13:05:18miranin fact i do :)
13:05:24miranfrom telegram: "I would appreciate if you would read it and report the mistakes and errors so I can correct it before going fully public. (That means - please do not share the link just yet on Nim forum, Reddit, etc.)"
13:05:37dom96okay, no worries
13:05:47dom96I saw the link in #crystal-lang lol
13:06:00FromGitter<Vindaar> @miran sorry I haven't sent another mail about it yet, I'll continue reading later today :)
13:06:14mirani would wait couple more days for error corrections before publishing it widely
13:06:26miran@vindaar no probs :)
13:06:41dom96miran: Where shall I submit corrections?
13:06:54dom96oh, issue tracker
13:07:10dom96Maybe I can just make a PR actually
13:07:15dom96Nice to see asciidoc :D
13:07:33mirandom96: either issue tracker, or to me privately (here / gitter / telegram), whichever is easier to you
13:08:22miranasciidoc is nice, but for example - i haven't seen the option to have multiple web-pages, whilst keeping TOC as it is
13:10:00dom96You could try hacking that in with a post-processing step written in Nim ;)
13:10:13miranhehe, true that :)
13:12:35*xkapastel joined #nim
13:16:43FromDiscord<2vg> @cschardt it's vuepress.
13:16:43FromDiscord<2vg> my wiki repo: https://github.com/2vg/Nim-World
13:26:30*sz0 joined #nim
13:27:54*xkapastel quit ()
13:28:24*xkapastel joined #nim
13:36:41*yglukhov_ joined #nim
13:38:27*yglukhov[i] quit (Ping timeout: 240 seconds)
13:46:52mirandom96: should i expect more changes coming in in your PR, or should i merge it as it is?
13:46:52*yglukhov_ quit (Read error: Connection reset by peer)
13:46:53*yglukhov[i] joined #nim
13:48:09dom96merge it
13:50:53*clyybber joined #nim
13:52:26PMunchAraq, do you have any idea why packedjson doesn't like to have a .kind statement generated through a macro?
13:53:14PMunchI just get: Error: attempting to call undeclared routine: 'kind'
13:53:29PMunchCopying the output of the macro into a program works just fine..
13:53:40AraqPMunch: 'kind' is a proc, newCall(bindSym"kind", ...) ?
13:53:52Araqit is not a field.
13:53:59PMunchIt's made with a quote do block..
13:54:11mirandom96: done. thanks for the correction!
13:54:46*yglukhov_ joined #nim
13:57:57*yglukhov[i] quit (Ping timeout: 240 seconds)
14:06:11PMunchThis is what the macro generates: http://ix.io/1cUL
14:06:21PMunchA call with kind as symbol and data as an ident
14:06:34PMunchdata is a JsonNode
14:07:46*yglukhov_ quit (Remote host closed the connection)
14:07:59AraqPMunch: try to generate it without 'quote do'
14:08:34Araqnot sure why quote do is so fragile, I blame the lack of .dirty support for the hidden template that is generated
14:10:25PMunchOoh, interesting
14:10:47PMunchlet kindSym = bindSym"kind" and then use `kindSym` instead of kind in the quote do block seems to work
14:14:00PMunchHmm, what was the difference between a JsonTree and a JsonNode again
14:14:12PMunchOne was mutable and the other is just a view into a larger tree?
14:16:58*Snircle joined #nim
14:20:31Araqexactly
14:22:45PMunchHmm, well that's a bit troublesome..
14:23:14PMunchHow do you generate a JObject in packedJson without the %* macro?
14:23:24Araqthe %* macro exists
14:23:42Araqand works as in json.nim
14:24:29PMunchYeah I know, but I'm generating this through a macro. So it was easier to just use result["key"] = value for each of the key value pairs
14:25:31PMunchWait.. That should work fine..
14:25:41Araqyou need to know the path
14:26:00Araqroot["field", 0, 1, "fieldB"] = newValue
14:26:10Araqas only the root node can be used to update fields
14:26:17Araqthis could be your problem
14:26:38PMunchAh, the problem was in my array construction
14:26:53PMunchI did result["key"].add for each field
14:27:01PMunchHmm
14:40:23*yglukhov[i] joined #nim
14:44:47*yglukhov[i] quit (Ping timeout: 245 seconds)
14:47:29*miran quit (Ping timeout: 248 seconds)
14:47:43Vladartrying out gouache and acrylics https://i.imgur.com/bZkw8LO.jpg
14:48:04Vladaroops, wrong channel, sorry
14:48:39dom96Beautiful though :)
14:48:53PMunchYeah, nice paintings :)
14:49:05Vladarthanks )
14:53:04*nsf quit (Quit: WeeChat 2.1)
14:59:30*yglukhov[i] joined #nim
14:59:36*yglukhov[i] quit (Remote host closed the connection)
14:59:51*yglukhov[i] joined #nim
15:01:01*clyybber quit (Quit: good night)
15:03:22*SenasOzys quit (Ping timeout: 264 seconds)
15:11:11FromGitter<Varriount> Miran: May I send you a list of grammatical suggestions later?
15:16:55dom96Varriount: Check PMs
15:19:53*find0x90_ joined #nim
15:21:48*yglukhov[i] quit (Read error: Connection reset by peer)
15:21:55*yglukhov[i] joined #nim
15:21:59*find0x90 quit (Ping timeout: 276 seconds)
15:24:12FromGitter<mratsim> (https://files.gitter.im/nim-lang/Nim/rh2t/2018-06-10_17-23-49.png)
15:24:29FromGitter<mratsim> ^Time machine in Nim devel?
15:25:29Yardanico1-year old PR :D
15:28:12*jjido joined #nim
15:28:38FromGitter<cschardt> @2vg: thanks!
15:31:39shashlick7zip is now wrapped - https://github.com/genotrance/nim7z
15:36:18Yardanicoshashlick, btw, I've used easylzma a while ago:
15:36:27Yardanicohttps://github.com/Yardanico/nim-osureplay/blob/master/src/lzma.nim
15:36:41Yardanico(it's just a wrapper over 7z )
15:37:03Yardanicoshashlick, thanks for the project though, it will definitely be very useful
15:40:02*edcragg quit (Quit: ZNC - http://znc.in)
15:40:19*edcragg joined #nim
15:45:51*SenasOzys joined #nim
15:51:38*PMunch quit (Quit: leaving)
15:52:54shashlickNice yardanico
15:53:29*find0x90_ quit (Quit: find0x90_)
15:53:40shashlickI had to add the ability to rename generated files to nimgen to wrap this, should be useful to varriount as well
15:54:23*SenasOzys quit (Ping timeout: 265 seconds)
15:59:44*Shivelight quit (Quit: Connection closed for inactivity)
16:00:55Yardanicoshashlick, your nimgen project is amazing
16:09:10*yglukhov_ joined #nim
16:12:21*yglukhov[i] quit (Ping timeout: 240 seconds)
16:17:19*jjido quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
16:20:01*yglukhov[i] joined #nim
16:21:47*yglukhov_ quit (Ping timeout: 276 seconds)
16:22:33*rauss quit (Read error: Connection reset by peer)
16:24:50*rauss joined #nim
16:24:57*yglukhov[i] quit (Ping timeout: 264 seconds)
16:30:53shashlick👍 thanks I'm glad it's useful, I just wonder whether the wrappers are being used or not
16:42:08*SenasOzys joined #nim
16:59:46FromGitter<data-man> @shashlick Now with pegs! :)
17:07:08FromGitter<ephja> my code is too accessible. I need more obscure unicode operators
17:10:26*elrood quit (Quit: Leaving)
17:13:49*yglukhov[i] joined #nim
17:18:05*yglukhov[i] quit (Ping timeout: 260 seconds)
17:26:34shashlickdata-man: yep, will have to slowly migrate to pegs
17:28:30FromGitter<ephja> `let camera = 𝕀3`
17:28:54Yardanicoshashlick, there's also pure-nim regex
17:31:37*Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
17:31:41FromGitter<ephja> `↻(not window.shouldClose):...`
17:32:29shashlickYa will mostly migrate to pegs though native regex might be easiest to do
17:34:25Yardanico@ephja this is easy to do in Nim :) just make a template probably
17:35:16FromGitter<ephja> that's what I did. the only problem is the difficulty of inputting various symbols
17:36:29shashlickI hope to make it seamless to use c libraries with nimgen, simple nimble install should work anywhere and underlying wrapper stuff should be invisible
17:36:51shashlickThat way, we don't need to rewrite every single library natively in Nim
17:37:25shashlickI'd rather we write a Nim API layer over the wrapper instead
17:38:15shashlickLike make it possible to nimble install pcre and not require any dlls
17:38:32shashlickCompile in the code directly
17:45:24*miran joined #nim
17:48:53Yardanicoshashlick, well, but we need to aim to have more pure-nim libraries in the future
17:49:20Yardanicoso we should probably write our libraries if we have time and resources :)
17:58:28AraqI still think a *simple* graphics library is the most important thing
17:58:51Araqsure plenty of other topics are not yet covered well by libraries
17:59:05Araqbut with graphics you can attract people
17:59:42Araqand graphics.nim wasn't bad, needs a port to SDL2
17:59:49shashlickI'm not a fan of rewriting, rather leverage as much as possible and make it seamless
18:00:16Araqjust imagine we have graphics examples like others have shitty "hello world" examples
18:00:19dom96hrm, anything in the stdlib to format large numbers nicely (with commas etc.)?
18:00:27dom961000 -> 1,000
18:00:39Araqstrformat? no idea if it supports it
18:01:02shashlickSomeone else is already done the hard work, why reimplement and waste that time, new bugs, etc
18:01:44shashlickAraq: are you saying wrapper or new lib?
18:02:22*yglukhov[i] joined #nim
18:03:34Araqneither.
18:03:45Araqport the existing graphics.nim to use SDL2 under the hood
18:03:50Araqand maybe rework its event handling
18:04:28Araqso it fits your goal, no wheel reinventions
18:06:29Araqshashlick: that said, wrappers tend to be mediocre because of the cross platform issues, you need to ship DLLs for 3 different OSes, 32 vs 64 bit builds etc etc
18:06:57*yglukhov[i] quit (Ping timeout: 256 seconds)
18:08:19Araqand few have an environment where they can develop for 3+ different OSes
18:08:34Yardanicomost of nimgen wrappers compile C in place the binary
18:08:48*Araq himself relies on travis to cover Linux testing...
18:08:58YardanicoAraq, you have us for Linux testing :D
18:09:25Araqsure but maybe I'll try to fix my linux VM ... eventually...
18:09:51Yardanicouse something simple like manjaro or mint :)
18:10:14Araqcould also dual boot but then I waste precious hard disk space that I instead need for all this porn...
18:11:24*SenasOzys quit (Remote host closed the connection)
18:13:00*SenasOzys joined #nim
18:15:04shashlickYep, most nimgen wrappers are compiled in so no dll dependencies
18:15:19Yardanicothis can probably be even done to SDL2
18:15:24Yardanico(but IDK)
18:15:36shashlickBut it needs to be smarter, run configure or cmake behind the scenes, right now it's more delicate
18:16:07AraqI usually modify the code to not require configure or cmake.
18:16:41Araqit's more work but makes for a less fragile build process
18:17:10Yardanicowell, it would require to manually update the wrapper to newer versions of libraries
18:17:31*nsf joined #nim
18:20:13FromGitter<mratsim> replace graphics by cross platform UI toolkit :P
18:20:39FromGitter<mratsim> imagine everyone complaining about Qt, Electron and GTK, and here we go with Nim-UI-Toolkit
18:21:05Yardaniconimx :P
18:21:11Yardanicobut it uses SDL
18:24:46shashlickRight now nimgen doesn't support configure, cmake etc, and only gcc
18:25:18shashlickAm thinking of adding configure and cmake and fall back if those aren't installed
18:26:08shashlickLike for nimarchive, I have it working on Windows and Linux but have to hard code a config.h. If I can detect these tools then can generate live instead
18:26:27FromGitter<mratsim> I think you’re in for a world of hurt but who knows
18:26:51shashlickAnyway, my main concern with nimgen as it stands is whether people are able to use the wrappers successfully
18:27:08shashlickI have received very limited feedback
18:27:35Araqmratsim: I agree on the native Nim UI toolkit but
18:27:47Araqa) It's much more work than a newer graphics.nim
18:27:56FromGitter<mratsim> (yeah of course)
18:28:01Araqb) Trustable is working on it, help him
18:29:05Araqalso data plotting only requires a graphics.nim, not a fullblown UI lib
18:29:26Araqunfortunately I lost the source code for my putpixel based plotting... :-)
18:29:39FromGitter<mratsim> for plotting this works - https://github.com/Vindaar/NeuralNetworkLiveDemo
18:30:29AraqI am not familiar with "Plotly", what is it based on?
18:30:48YardanicoJS
18:30:51FromGitter<mratsim> d3.js
18:30:58Araqlol ok
18:31:12FromGitter<mratsim> it’s taking data science by storm and replacing matplotlib in Python
18:31:22Araqnot even a bad solution, compared to gnuplotlib and matplotlib
18:32:23Araqmatplotlib -- one week trying to produce scatter plots that show something.
18:33:02Araqputpixel -- 4 hours to reimplement scatter plots with transparency so that you could see something. :P
18:33:10miranif MPL is too complicated, try seaborn
18:33:16federico3pandas.
18:34:44YardanicoNimData :P
18:34:55FromGitter<data-man> @shashlick TinyCC supports -E too. You can try static links it to nimgen and don't use gcc. We have the wrapper. Just an idea. :)
18:36:43FromGitter<ephja> ```code paste, see link``` ⏎ ⏎ hmmm [https://gitter.im/nim-lang/Nim?at=5b1d6fbb0b75bc7d5afb22c3]
18:36:50miranfederico3: yup, pandas' plot wrapper is also quite nice
18:37:01miranwell, everything is nice compared to MPL :P
18:37:33Araqgood to know I'm not up to date and people stopped with matplotlib :P
18:40:27miranunfortunately, when you want something special, it is most likely that you'll need MPL. but for some basic/typical stuff - there are these much saner wrappers around it
18:42:14shashlickdata-man: but you need to run pre processing thru the same compiler that builds Nim as well
18:42:27shashlickThe Nim code that is
18:42:42shashlickElse your idea may not line up
18:42:51shashlickIfdef, not idea
18:44:47FromGitter<mratsim> for pandas plotting I like cufflinks
18:45:00shashlickThe main thing is that bigger libraries use autoconf or similar methods to discover what's supported by the compiler
18:45:24shashlickUntil nimgen abstracts that, it's not going to scale for more complicated libs
18:45:32FromGitter<data-man> @shashlick Maybe tcc can use gcc-headers, but I haven't tried it.
18:46:20shashlickWell I've not heard from anyone asking for clang or vcc support yet so haven't bothered
18:47:21shashlickAnd regardless, anyone installing nimgen has already installed nimble, Nim and hence a compiler
18:52:18*jjido joined #nim
18:53:29Araqshashlick: maybe add yet-another-build-tool to nimgen?
18:55:01Araqor generate nimble tasks that run ./configure
18:56:17*yglukhov[i] joined #nim
18:56:33*lainon joined #nim
18:58:12*yglukhov[i] quit (Read error: Connection reset by peer)
18:58:49*yglukhov[i] joined #nim
19:00:33FromGitter<ephja> ok so `foo(bar=baz)` where 'bar' is misspelled causes that
19:03:15shashlickAraq: is that a real thing? yet another build tool?
19:03:36shashlickI was just going to detect and call cmake or configure
19:03:38Araqwell I'm just thinking aloud
19:03:55shashlickSince that's what most c libraries use
19:03:58FromGitter<ephja> Boost.Jam!
19:04:09FromGitter<data-man> tup :)
19:04:27FromGitter<kayabaNerve> I hate Make
19:04:29FromGitter<kayabaNerve> I love Nake
19:06:14Yardanicoyou should use nimscript :)
19:06:37FromGitter<kayabaNerve> Sad to hear dom96 say it's been deprecated. Happy to see the repo doesn't agree, even if dev has died down
19:07:01Araqnimscript was inspired by nake fwiw
19:07:12FromGitter<kayabaNerve> Yardanico: Blasphemer!
19:07:31Araqbut nake handles some things better, admittedly
19:07:44YardanicoI mean that nake isn't required in most cases
19:08:30FromGitter<kayabaNerve> Dom also said that :P It looks good. It just isn't as powerful and I liked the concept of the full Nim language being the build tool, not a subset.
19:08:32shashlickWell but at least in nimgen case, I don't need to create a new build system, just invoke the build system chosen by the underlying c lib
19:08:48Yardanico@kayabaNerve nimscript allows you to use almost all Nim modules :)
19:08:53Yardanicoand a lot of third-party ones
19:32:24Araqwell that's debatable
19:32:53Araqand in fact, nimscript is a historical accident, I wouldn't write a VM for Nim today
19:33:20Araqbut it's pretty cool and works and should be useful to embed in game engines
19:34:33YardanicoAraq, what would you do instead? :)
19:34:44FromGitter<kayabaNerve> Yardanico Not Nimscript
19:34:47FromGitter<kayabaNerve> He just said that
19:34:47Yardanicofor macros/compiletime stuff/etc
19:41:44*sz0 quit (Quit: Connection closed for inactivity)
19:45:55*jjido quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
19:47:10AraqYardanico: compile Nim snippets to native code and execProcess() it
19:49:49YardanicoAraq, same with macros?
19:50:18*thomasross quit (Read error: Connection reset by peer)
19:50:37Araqthat's how I would implement the macro system, yeah
19:51:01Araqbut that only works with the compilation cache
19:52:26Araqand in fact, it's not 100% sure this would work.
19:52:39Araqit's a research project for later Nim versions
19:55:36*Arrrr joined #nim
19:56:11*Vladar quit (Quit: Leaving)
20:00:33*Trustable joined #nim
20:02:30FromGitter<notTito> hey speaking of macros, here is a dump question: Suppose i have a local variable, how I check where its used in the body? Compare the NimNodes directly with `==` use eqIdent or store the .ident of that variable and check?
20:03:39FromGitter<krux02> @notTito do not use the .ident member of a node anymore
20:03:51FromGitter<krux02> the type NimIdent is deprecated
20:04:31FromGitter<krux02> comparing nimnodes with `==` should be fine
20:04:52*PMunch joined #nim
20:04:55FromGitter<krux02> when the AST has the same symbol several times they should be identical
20:05:13FromGitter<notTito> ok thanks krux02
20:05:33FromGitter<krux02> so you should have a typechecked AST, and then to a recursive traversal and test for equality.
20:06:23FromGitter<krux02> do the typechecking because, then you can be shure that another symbol with the same name does not match.
20:13:25*chrismc joined #nim
20:13:28FromGitter<notTito> makes sense but in my case I cant do that bc i was experimenting with a `this` macro
20:14:06FromGitter<notTito> not sure how its going to work or if its even possible
20:17:22*chrismc quit (Client Quit)
20:17:40*chrismc joined #nim
20:18:50*chrismc quit (Client Quit)
20:18:53*chrismc_ joined #nim
20:19:34*chrismc_ quit (Client Quit)
20:19:56*chrismc joined #nim
20:26:53FromGitter<turbolent> Hi there! I'm trying to get Nim working on my rusty old PowerMac G5 on Mac OS X 10.4. With https://github.com/nim-lang/Nim/compare/devel...turbolent:macosx-powerpc I get the C sources of v0.18 compiling , and the nim command spits out its help. However, compiling a simple hello world fails with: ⏎ ⏎ ```code paste, see link``` ⏎ ⏎ Any idea what could be wrong here? [h
20:26:53FromGitter... ttps://gitter.im/nim-lang/Nim?at=5b1d898cddd61d08e545a67d]
20:30:08ArrrrIt must be that, until now, no one implemented that proc in mac i guess
20:30:27Arrrror for you exact os
20:33:03FromGitter<turbolent> how come the nim works on recent versions of macOS though? I assumed the implementation at the very end in lib/system/excpt.nim was picked. is that used for other mac platforms/versions, or another one? it looks like this should work just fine even on the old PPC
20:34:14ArrrrThat's a good question. Nim figures it out through defines. So there must be something special about your OS.
20:35:29Arrrrhttps://github.com/nim-lang/Nim/blob/devel/lib/system/platforms.nim
20:39:20Araqturbolent: check instead how far you get with 'koch boot --gc:none'
20:39:37AraqI think Power32 needs adjustments in its stack handling
20:39:50Araqand a wrong GC can sadly produce totally misleading Nim compiler errors
20:41:22Araqthe responsible file would be system/gc_common.nim
20:42:01FromGitter<turbolent> On the PPC target machine or on my macOS/x86 host? On the target, I only got `build.sh` succeeding, `./bin/nim c koch` fails with the same error, as it does for the hello world program
20:42:14FromGitter<turbolent> interesting
20:43:08Araqwell you cross compile on the host for the target machine
20:43:20Araqand compiling on the host should work
20:43:31FromGitter<turbolent> I see, will give that a shot
20:43:39AraqI assumed you reported this result when the compiler already runs on the PPC target machine
20:44:16FromGitter<turbolent> yes, report was from the PPC target
20:46:28Araqalternatively build with --gc:markAndSweep
20:47:01Araqthe M&S GC probes the stack so rarely that it doesn't matter much if the stack scanning is 100% correct :P
20:47:53shashlickdoesn't look like our cached copy of mingw contains what's required to run cmake or autoconf on windows
20:48:18*chrismc quit (Quit: Leaving)
20:48:27*chrismc joined #nim
20:50:00Araqautoconf on windows never worked for me
20:50:15Araqand on other OSes it rarely worked
20:50:20shashlickcmake ships a separate installer/.zip
20:50:34shashlickconfigure has worked fine for me in msys but that's a huge dependency
20:52:18*chrismc quit (Client Quit)
20:52:29*chrismc joined #nim
20:53:06*chrismc quit (Client Quit)
20:53:17*chrismc joined #nim
20:54:03*chrismc left #nim (#nim)
20:56:37FromGitter<zetashift> @Yardanico is it bad that nimx is based on sdl?
20:56:50*chrismc joined #nim
20:58:12shashlickcan you make a PR that's based on another PR?
20:58:29shashlickwithout reusing the branch of the first PR
21:04:46Yardanico@zetashift not really
21:08:05chrismcQuick question, I'm pretty new to Nim and I'm getting a "SIGSEGV: Illegal storage access. (Attempt to read from nil?)" error, however it doesn't indicate the line for the error. How do I determine which line has the issue?
21:09:17shashlickwithout reusing the branch of the first PR
21:09:31Yardanicochrismc, are you getting this error at compile-time or runtime?
21:09:42Yardanicoand if at runtime, did you compile your program in release mode?
21:10:01*nsf quit (Quit: WeeChat 2.1)
21:10:22chrismccompile time, compiled with just "nim c -r"
21:10:25Araqshashlick: I don't know
21:10:45Araqchrismc: don't use types as values and vice versa
21:11:00shashlickok, guess I have to wait for my PRs to be integrated before working on anything building on top of them
21:11:13Yardanicochrismc, it's an error in the compiler then, can you post source code on gist or somewhere else? if it's not confidential
21:11:41Araqshashlick: actually, you only have to wait with *pushing* more commits onto your PR branch
21:12:37chrismcYardanico: https://gitlab.com/chrismc/fooling-around-with-nim/blob/master/datastructures/stack.nim
21:12:47shashlickya but don't want to go too far into the rabbit hole if I receive feedback which changes the design too much
21:13:25FromGitter<Varriount> Shashlick: What pull request is this?
21:13:59shashlickhttps://github.com/dom96/choosenim/pull/63
21:21:24Yardanicochrismc, it doesn't crash for me on devel
21:21:31chrismcinteresting
21:21:32Yardanicochrismc, and you have a typo on line 99
21:21:38Yardanicochrismc, stack -> stack1
21:21:46Yardanicoand it passed all tests
21:22:13chrismcyea, that's the error, shit, thanks Yardanico for the 2nd look
21:22:26chrismcwhen fixed to "stack1", it runs fine
21:22:45*dddddd quit (Ping timeout: 268 seconds)
21:22:47Yardanicochrismc, compiler crashing is always considered a bug, but it doesn't happen on devel, so it's fixed :)
21:23:02Yardanicocompiler gave me a clear error "test.nim(99, 17) Error: undeclared identifier: 'stack'"
21:23:26Yardanicochrismc, what nim version are you using?
21:23:34Yardaniconim -v
21:23:44chrismc0.18.0
21:24:12Yardanicovery strange, it doesn't even crash with SIGSEGV on play.nim-lang.org
21:24:29chrismcinteresting, I wonder what's happening there
21:24:45*euinix joined #nim
21:25:10*euinix quit (Client Quit)
21:27:31*Arrrr quit (Quit: Arrrr)
21:28:18*dddddd joined #nim
21:29:32*yglukhov[i] quit (Ping timeout: 268 seconds)
21:30:52chrismcI see what happened: I declared a type "Stack" and then used "stack.pop", which basically was me "using a type as a value" per Araq's message above
21:33:37chrismcdoes that make sense? I'm pretty new with Nim
21:33:57*yglukhov[i] joined #nim
21:36:15shashlickcan you print stuff in the `=destroy` proc? nothing prints out, how do I know it is running
21:36:55Yardanicochrismc, maybe you've used "Stack.pop" ?
21:38:02chrismcI don't know what happened, but when I use "discard stack.pop" in my file, it does the error message I saw before without any compile errors
21:38:05FromGitter<ephja> if it doesn't print anything then it's not being called
21:43:01shashlickok then I need a review of my SvnzFile object and destructor: https://github.com/genotrance/nim7z/blob/master/nim7z.nim
21:43:07shashlickappreciate it
21:43:20chrismcYardanico: When I run that same code with the devel branch, it still gives me that same output with no "undeclared identifier" message
21:44:13Yardanicochrismc, that's veeeeryy strange
21:44:18Yardanicocan you post the exact code?
21:44:23Yardanicoare you sure it's the same as on gitlab?
21:46:04chrismcHere is a static snippet: https://gitlab.com/snippets/1723320
21:46:26chrismcthe last line "discard stack.pop" does the SIGSEGV
21:47:10chrismcit does it on stable and devel branches for me
21:48:03Yardanicowell, lol, it still gives "discard stack.pop" to me
21:48:10*Trustable quit (Remote host closed the connection)
21:48:12YardanicoI mean "test.nim(132, 11) Error: undeclared identifier: 'stack"
21:48:15*miran quit (Quit: Konversation terminated!)
21:48:36chrismcyea, weird, it gives me that in play.nim-lang.org, but on my local PC nothing
21:52:13chrismcCan I do anything else to troubleshoot this? I'd like it if my nim would give me the same errors as everyone else, lol
21:53:02Araqwhat's the OS? how did you install Nim? what does 'nim -v' really say
21:53:44chrismcLinux and Nim Compiler Version 0.18.0 [Linux: amd64]
21:53:44chrismcCopyright (c) 2006-2018 by Andreas Rumpf
21:53:44chrismcgit hash: 855956bf617f68ac0be3717329e9e1181e5dc0c6
21:53:44chrismcactive boot switches: -d:release
21:54:00FromGitter<turbolent> @Araq same error with `./koch csources --debuginfo -d:debug --linedir:on --debugger:native --gc:none`
21:55:00Araqturbolent: also try -d:useSysAssert -d:useGcAssert but first learn how to cross compile without 'koch csources
21:55:14Araqwhich is super slow as it does way too much work
21:55:24chrismcAraq: I can file a bug report, if it is easier to gather info that way
21:55:35*Cthalupa joined #nim
21:56:03Araqok but don't expect miracles if nobody can reproduce your problem ;-)
21:56:47FromGitter<turbolent> @Araq thanks, I'll try that and also look into actual cross compiling, but I doubt that is even possible
21:56:51Araqnim c --cpu:power32 --os:macosx --genscript --nimcache:osxpower32 compiler/nim.nim
21:57:11chrismcyea, no worries, there are more important things for you to work on of course, I'll file a ticket with everything I know so far
21:57:12Araqand then move the osxpower32 dir to your machine and run the produced sh build script
21:57:48Araqis much faster for cross compiling
21:58:57*CodeVance joined #nim
21:59:50FromGitter<turbolent> @Araq great, thanks!
22:00:52Araqand thanks to shashlick we know --genscript works. right?
22:01:04CodeVanceThank you Araq
22:01:16shashlick👍
22:02:37AraqCodeVance: for what? :-)
22:02:51CodeVanceAraqyou can't "kill" a thread, the posix APIs pretty much deprecated it06:45:06Araqyou thread needs to support receiving a "shutdown" message06:45:14Araqor else you use a process instead.
22:03:06CodeVanceFor telling me to use a process instead
22:03:12Araqok lol
22:03:15Araqyou're welcome
22:04:28CodeVanceI have a problem when I try to load a nim dll that has an infinite loop in it. it blocks the main program and I don't know how to cancel loading it. For now I made a seperate program loaded the dll and checked if the program exited normally. is that correct?
22:05:09Araqpretty much the best solution afaict
22:06:34CodeVanceOk. its not a nim problem its a microsoft problem I think
22:06:41CodeVanceso nothing else I can do
22:06:47Araqor you wait until https://github.com/nim-lang/Nim/blob/araq-incremental-compilation/tests/compilerapi/tcompilerapi.nim lands in devel and use NimScript for your plugin system
22:06:57FromGitter<ephja> the loader actually doesn't detect loops?
22:07:21Araqdepends on whether Nim's VM is good enough for you.
22:07:38chrismcfor my weird problem, feel free to ask for more information there, low priority since it seemingly only affects me: https://github.com/nim-lang/Nim/issues/8011
22:07:50Araqthe VM can be made to not run into endless loops
22:09:18CodeVanceVM would be the best solution. I'll switch to it when its available?
22:09:55Araqneed to fix 3 test cases and then it'll be in devel. check tomorrow
22:10:02FromGitter<ephja> oh the vm
22:10:21Araqdon't you dare say anything against my vm :P
22:10:30Araqit's come a long way.
22:11:12Araqand it's a super cool register based VM
22:11:21Araqinspired by Lua's.
22:11:32CodeVanceis the vm like a full nim interpreter?
22:12:04FromGitter<ephja> I just read "dll"
22:12:07CodeVanceCan it do everything normal programs can
22:12:21Araqit's the only Nim interpreter
22:12:35Araqand it can't run the FFI ;-)
22:15:03krux02Araq: do you plan to kill the interpreter?
22:15:21Araqwithin the next decade, maybe
22:15:41*PMunch quit (Quit: leaving)
22:15:52Araqbut it's useful for game engines too.
22:16:03Araqso it probably should become just another backend
22:16:06FromGitter<ephja> I assume we don't have enough linalg libraries yet
22:16:20Araqand the macro system should be based on a plugin system.
22:16:28Araqbut that's far away.
22:16:38CodeVanceI wanted to experiment with hotloading nim libs
22:17:10krux02well I don't think it's wise to kill it. There are macros that are only used library internal, there is little point in compiling those scripts into a plugin and then execute them once
22:17:12FromGitter<ephja> plugin sandboxing
22:18:06krux02don't get me wrong, I like the idea of plugins, I think the nim to glsl compiler works much better as a Nim plugin than it does now
22:18:15Araqkrux02: if the caching works and the compiler is based on packed data structures it would be a performance win
22:18:28krux02but I don't think plugins are superior to the scripting system
22:18:37Araqand when was the last time you compiled a program/snippet only "once" anyway
22:18:59krux02isn't caching completely orthogonal to the macros as plugin transition?
22:19:14krux02Araq: the glm library does that
22:19:25krux02it generates all swizzle procs with a macro
22:19:28FromGitter<ephja> quake 3's vm loads bytecode produced by LCC
22:19:46krux02it's not used once, the macro is used 3 times
22:19:51Araqephja. pretty crazy, but if it worked for them
22:20:12krux02for the set "xyzw", "abcd" and "uv.."
22:20:16krux02I forgot the last part
22:20:30krux02but that's it, the macro isn't used outside of the library
22:20:45FromGitter<ephja> certain operations are not allowed. pretty sure that's how it worked
22:20:57krux02waht is LCC again?
22:21:14Araqa C compiler.
22:21:22Araqpretty dead these days.
22:21:55Araqwas based on a book about how to build table based code generators.
22:22:15Araqwas quite impressive in its days.
22:22:46Araqand small too. but it never produced code on par with GCC
22:23:45krux02I mean projects that are written for teaching purpose are often not rich in featuers like optimizations
22:23:54Araqwas then developed further by a guy who didn't understand C++ and wanted to improve C in his own silly ways, like so many other "C improvements"
22:23:55krux02but yea, I don't know much about it
22:24:34Araq"C++ is bad because '=' can mean more than a memcpy and so I added big nums as builtins to C..."
22:24:41krux02Well to be honest, c++ also has silly ideas, because people didn't know better
22:25:18Araqthe C++ haters usually fail to mention its silly ideas. :P
22:25:29krux02yea that is true
22:25:41Araqand instead come up with FUD.
22:25:54krux02the c++ haters usually are not able to properly program in c++ and don't like it becaues of it
22:26:13krux02but spotting the concrete problem is actually quite hard if you don't understand the language
22:26:16Araqoh it's fine to not be able to program in it.
22:26:40krux02Nim also has silly ideas
22:26:55Araqhuh, blasphemy.
22:26:58Araqlike what?
22:27:10krux02the "range" type
22:27:17krux02:D
22:27:44Araqyeah, that thing that helps Ada's proof engines surely was a mistake to copy over
22:27:51CodeVancelol I don't know if I've ever used the range type
22:27:57Araq:P
22:28:15krux02yes exactly
22:28:41krux02I just don't use the range types, and I am 100% happy with the language
22:28:52FromGitter<ephja> CodeVance: get out!
22:28:57FromGitter<ephja> :-)
22:29:48CodeVanceephja nim's pretty big. Maybe I've used them I just don't know???
22:29:53krux02well I like the idea to be able to specify constraints on numbers, I just don't think it should be builtin to the language
22:30:14krux02it's a type that has two static numbers baked into them
22:30:47dom96range types may have a niche use, but for that use they are great
22:30:59CodeVancerange for numbers? that's useful
22:31:14CodeVanceits not like its bothering anyone
22:31:36CodeVancedoes it look like this range[0..5]
22:32:05krux02and I am always confused it an expression is a type or a value
22:32:18krux02is 0 .. 5 a value or is it a type
22:32:21krux02I never know
22:32:43*yglukhov[i] quit (Ping timeout: 256 seconds)
22:33:02krux02and when 0 .. 5 is a range value
22:33:51krux02and range[0..5] is a rangle type that as a static range value as part of it, then why can't the range type be just a pure library feature?
22:34:00krux02a type with two static integer fields
22:34:05krux02but we had this discussion
22:34:10krux02it's not going anywhere
22:34:14FromGitter<ephja> it constructs a slice. maybe range[T:static[Slice]] would have been more obvious
22:34:17krux02so I won't continue here
22:34:41krux02Araq: I have a question to the compilation cache, because I don't quite understand the technical challenges with it
22:34:58krux02you mentiond the gathering technique
22:35:06Araqno, zahary did
22:35:14Araqbut go on
22:35:15krux02ah, sorry
22:35:54*CodeVance left #nim (#nim)
22:36:00krux02well yes, that technique is something that I would like to take advantage of, but I didn't even do it, because I didn't find a way that I was happy with
22:36:37krux02all I was able to do was to write macros that appended to this shared global state, and then the user was responsible to call another macro that then used this shared global state
22:36:38*CodeVance joined #nim
22:36:44FromGitter<ephja> I don't think you can implement it in Nim yet
22:36:52krux02but I din't find a way to take this resposibility away from the user
22:37:03krux02so I didched that idea
22:37:15krux02but I would really liket ot have it back.
22:37:41krux02so the question is, when there is a shared global state of something, what is the problem there with the compilation cache?
22:38:06Araqyou are asking about the 'sealing' step
22:38:15Araqproblem is that Nim lacks it.
22:38:38Araqand a compilation cache is the wrong thing to latch this feature onto.
22:38:48krux02what is the 'sealing' step?
22:39:19Araqwhat you described, what current needs to be done as 'proc callThisWhenDone() {.compileTime.}'
22:39:32Araqwhich everybody seeks to avoid :P
22:39:44krux02I am not saying I want to "latch" this feature on the compilation cache, but I would like to have that featuer, and I would like the compilation cache to work. σo I would like to know about where those two things bite each other.
22:40:02AraqI only used this technique in 'nimsqueak'
22:40:30krux02nimsqueak?
22:40:36Araqand often the lookup tables are moved to the runtime due to a lack of a sealing step
22:40:39krux02reminds of of newsqueak
22:40:51Araqwhatever, it's not important.
22:41:06Araqjester uses runtime routing tables for this reason
22:41:32Araq(and maybe because the VM doesn't support these tables with regexes in them properly)
22:42:09Araqwell. we discussed this here in #nim
22:42:20Araqand there are not too big obstacles to do this
22:42:59krux02and then?
22:43:20Araqbut the phase orderings would probably need to be changed, affecting everything
22:43:40Araqzahary claimed "it's not much work" but he always says that :P
22:43:49krux02I was actually thinking about making regular expressions available on the VM, as a reald binding
22:44:17AraqI am not convinced we need to change the phase orderings to get this feature
22:44:34krux02yea, I used to say that too. Wisedom told me, it is always a lot of work when you want to do it properly
22:45:41krux02it always, idea, concrete concept, implementation, problems, implementation, problems, done, works, bugs, fixing bugs, bugs, fixing bug, more bugs. Finally done, last bug ...
22:46:06krux02It just takes a long time to write something that works without problems when it is something non trivial
22:47:08krux02well back to topic, what do you/he mean with phase orderings?
22:47:12Araqwell partly that's because we made the mistake of making Nim not 100% "introspectable"
22:47:48Araqbut we'll get there, it's my "vision"
22:48:03krux02what do you mean with introspectable?
22:48:18Araqkrux02: zahary argued the sealing step needs to be done after sem but before transf
22:48:33krux02transf?
22:48:42Araqand it runs after the full program has been sem'checked
22:48:54Araqand before anything was transf'ed.
22:48:59Araqsorry, compiler terms.
22:49:05Araqthe compiler is a pipeline
22:49:12krux02well, I have to look that term up
22:49:16Araq'transf' is the first codegen step
22:49:22Araqit inlines iterators, for example
22:49:35krux02yea google corrects me
22:49:37Araqso the AST is not recognizable afterwards for macros
22:50:12Araqand we would need to change how this pipeline is setup.
22:50:25krux02in typed macro arguments, iterators are already inlined
22:50:34Araqyeah :P
22:51:11Araqthat's why I think it's not even important ;-)
22:51:22Araqmacros already need to be able to eat this shit
22:52:18krux02I even wrote a pattern matcher for an inlined iterator to compile it to glsl, because glsl does not have a goto and I can't compile te inlined iterator with the named break statements
22:52:30Araqhave you seen my packedjson?
22:52:38krux02no
22:52:58Araqwell I long for using this idea for Nim's ASTs
22:52:58krux02link?
22:53:10Araqhttps://github.com/Araq/packedjson
22:54:00FromDiscord<jos> QUICK QUESTION
22:54:09FromDiscord<jos> best way to install nim for an ephemeral CI build?
22:54:13FromDiscord<jos> best way to install nim for an ephemeral CI build?
22:54:33Araqcheck Nim's .travis and appveyor scripts
22:54:38krux02well that is actually a great idea, I never thought about that it might be possible for the ast.
22:54:45FromDiscord<jos> ok
22:54:46FromDiscord<jos> thanks
22:54:47Araqit is possible.
22:55:05Araqand it creates an aliasing free AST.
22:55:15Araqwith a O(1) destructor.
22:55:41Araqthat you can copyMem() to a different thread or write to disk.
22:56:19krux02I think the write to disk is pretty useful for compilation caching
22:56:29Araqoh really, tell me about it :P
22:57:30FromDiscord<jos> i looked in a third party repo and they use a script in the nim repo called ci/build.sh
22:57:44Araqyou check the input file length, you know the AST is usually 80% of that size, you allocate the buffer for the complete AST in one alloc
22:57:51krux0294MB compared to 1.2GB is pretty neat
22:58:28Araqthe input file is 180MB
22:58:30krux02(comment about packed jsone)
22:58:55krux02loading 180MB can be pretty fast
23:00:43krux02Araq: I think stdlib json should become deprecated and people should be encouraged to use packedjson instead. But thats just my opinion.
23:01:07Araqthe API is harder to work with though and I cannot support all of json.nim
23:01:37krux02what is missing?
23:01:58Araqwell you can't change the tree from a child node
23:02:09Araqonly from the root node (called JsonTree)
23:02:22Araqbecause the aliasing otherwise becomes unmanagable
23:02:34krux02that is of course true
23:02:44Araqyou could introduce COW and versioning perhaps, but I didn't need it
23:03:04krux02COW?
23:03:41Araqcopy-on-write
23:05:49*chrismc quit (Quit: Leaving)
23:06:00krux02well I am thinking about a reference to a node that knows all parents, and when you change the node by adding more stuff to it, it krows the space the node has like a seq does, and then you can just write to it, but at a certain amount it has to reallocate everything, the entire tree, so everything stays flat.
23:06:37krux02or the nods that you edit, are replaced in the tree by some sort of escaped pointers
23:07:04krux02so that the json is flat except the edited nodes
23:07:11krux02but yea very hypothetical
23:07:36krux02I only use json for parsing and for serialization, I neverd needed mutable json.
23:08:31krux02Araq: macros generate ref based Nodes
23:08:46krux02so they explicitly allow reuse of Nodes
23:09:07krux02do you want to give Copy semantic to all Nim Nodes?
23:10:49Araqyeah
23:11:16krux02would break a lot.
23:11:35krux02I am not the one who is against breaking a lot of code when it is eventually changing for the good.
23:11:47AraqI know but you can also use the packed AST in the compiler
23:11:52Araqand the ref based AST in the VM
23:12:00Araqand convert between them
23:12:05krux02ok
23:13:02krux02then I won't need to search for all cases where I append to a node after I've inserted it into another node
23:13:35krux02Well it's getting late again, and I need to get up tomorrow again, so I would postphone this discussion for tomorrow. I will be online the entire day starting from 10:00
23:13:43krux02until maybe 20
23:13:50krux02so I have to say good night
23:14:00Araqok, bye
23:19:55*krux02 quit (Quit: Leaving)
23:38:53*xet7 quit (Remote host closed the connection)
23:52:21FromGitter<turbolent> ```$ ./hello ⏎ Hello from PowerPC G5 on Mac OS X 10.4``` [https://gitter.im/nim-lang/Nim?at=5b1db9b5dd54362753008a2e]
23:56:02FromGitter<turbolent> compiling the compiler directly sped up iteration a lot, but no matter what I tried, every GC and option resulted in the original error. once I commented out the `setControlCHook` forward declaration in lib/system.nim:3216 it worked. Really nice to finally have Nim working, but I still wonder why the implementation in lin/system/excpt.nim isn't found and used 🤔
23:59:36*BitPuffin quit (Remote host closed the connection)