<< 01-02-2019 >>
00:00:02*ghost64 quit (Quit: See you!)
00:00:52*ghost64 joined #nim
00:05:22dom96gc_safe is for parallelism
00:05:40*krux02 quit (Remote host closed the connection)
00:05:58dom96if you can't alloc then the async macro might not give you much
00:06:09rayman22201yes, but part of that is preventing heap allocations correct?
00:06:41dom96not really, it's about ensuring that heap-allocated data on one thread is not accessed by another
00:07:12dom96GC safe means "this proc does not access any global variable/state that may be owned by another thread or call any functions that are not gc safe"
00:07:14dom96'night
00:08:59rayman22201fair enough. gn
00:10:10*stefanos82 quit (Remote host closed the connection)
00:13:19FromDiscord_<exelotl> ahh ok this won't work because yield in valid nim code always needs a value. I guess I'll need to come up with my own magic word
00:28:34*Cthalupa quit (Ping timeout: 246 seconds)
00:29:12*Cthalupa joined #nim
00:41:58*zachk quit (Quit: Leaving)
00:47:42*seni quit (Ping timeout: 246 seconds)
01:21:45*libman_ is now known as libman
01:27:19FromDiscord_<exelotl> god it feels like I'm writing a mini compiler at this point. I got if statements working but sometimes they generate unnecessary code. Tomorrow I guess I'll tackle loops or something
01:41:28*Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
01:50:43FromGitter<zacharycarter> omg tech interview over
01:50:47FromGitter<zacharycarter> thank goodness
01:51:03FromGitter<zacharycarter> so exhausted and stressed - can finally relax
01:57:45FromGitter<zacharycarter> 70-75% confident in my effort(s) - I didn't do perfectly but I think I did well
01:57:49FromGitter<zacharycarter> or well enough I guess
01:57:57FromGitter<zacharycarter> hopefully will find out tomorrow :D
02:01:20*ng0 quit (Quit: Alexa, when is the end of world?)
02:36:04*radgeRayden joined #nim
02:45:14*vlad1777d quit (Ping timeout: 250 seconds)
03:01:51*Tyresc quit (Quit: WeeChat 2.4-dev)
03:02:24*banc quit (Quit: Bye)
03:25:53*banc joined #nim
03:37:37FromGitter<gogolxdong> How to cast ⏎ ⏎ ```type ⏎ EventHandler* = proc (ev: Event; target: VNode) {.closure.}``` [https://gitter.im/nim-lang/Nim?at=5c53bf01f04ef00644bd6f3c]
03:40:15*darithorn quit (Remote host closed the connection)
04:03:23*nsf joined #nim
04:07:10FromGitter<Varriount> @zacharycarter Looking at a new job?
04:16:24*noonien quit (Quit: Connection closed for inactivity)
04:22:11FromGitter<zacharycarter> @Varriount - just finished 3rd round with Riot Games
04:22:21FromGitter<zacharycarter> next is on-site
04:26:14FromGitter<zacharycarter> assuming I made it past 3rd round - but like I said earlier - I think I did well enough
04:26:20FromGitter<zacharycarter> but tomorrow will tell :P
04:47:32*westerbly_ joined #nim
04:49:39*radgeRayden quit (Ping timeout: 250 seconds)
04:55:36*westerbly_ quit (Remote host closed the connection)
04:56:44*mbomba joined #nim
04:57:01*radgeRayden joined #nim
04:57:39FromGitter<Varriount> @zacharycarter Cool! I definitely think you're a good match for a game company.
04:58:26*mbomba quit (Client Quit)
05:00:37FromGitter<zacharycarter> @Varriount thanks mang! me too! :D
05:01:08FromGitter<zacharycarter> I learned so much in the past 1.5 weeks - learned most common algo STC and same with ops on DS's
05:01:32FromGitter<zacharycarter> also understand how most of them work - I don't know all algos yet obv / how to solve all problems efficiently - but I could talk to those points when making my solutions
05:02:16FromGitter<zacharycarter> I also learned about converting from decimal to binary and vice versa - which I had learned previously but not in as memorable as a way
05:03:49FromGitter<zacharycarter> also learned about different C++ things - like how virtual functions work with a vtable and vptr and also neat pointer arithmetic things
05:04:07*NimBot joined #nim
05:04:35FromGitter<zacharycarter> so many effing things to cover and learn - plus math skills on top of it - for a non-CS person it's a lot, but I feel like even if I don't get this gig, I'll be well prepared for whatever I pursue next
05:06:50*dddddd quit (Remote host closed the connection)
05:07:30FromGitter<zacharycarter> connecting practical experience w/ understanding is addicting - like I've used bit flags / sets and masks for so long
05:07:40*radgeRayden quit (Remote host closed the connection)
05:07:43FromGitter<zacharycarter> but now that I understand them - I feel like I have super powers or something :P
05:08:11*radgeRayden joined #nim
05:11:15FromGitter<zacharycarter> http://tiny.cc/jvxu2y - was a godsend
05:11:32FromGitter<zacharycarter> plus google
05:11:54FromGitter<bung87> I’m curious about what’s the job title?
05:12:11FromGitter<zacharycarter> I'd be a game programmer on the game modes team for League of Legends
05:12:24FromGitter<zacharycarter> so introducing new game play modes to the game
05:13:13FromGitter<bung87> so you will programing LOL game server?
05:13:45FromGitter<zacharycarter> I doubt it's much networking code - I mean I'm sure I'd have to work with code that is running in a multiplayer capacity
05:13:55FromGitter<zacharycarter> but I have a feeling most of the plumbing will be in place to handle that gracefully
05:14:15FromGitter<zacharycarter> I think it's more about implementing gameplay logic - like if a new game mode is designed or introduced, how to implement it best
05:15:09FromGitter<bung87> sounds cool ,whether the frontend or backend
05:15:26FromGitter<zacharycarter> thanks! my dream job is to work on a game engine - but I need to get into the industry first
05:15:30FromGitter<zacharycarter> and I think I could land at much worse spots than Riot
05:15:46FromGitter<zacharycarter> so - this is an amazing opportunity for me, I just hope I didn't bungle it haha
05:15:50FromGitter<zacharycarter> but either way - it was a great experience
05:18:00FromGitter<bung87> yeah ,definitely. if you have art work experience,you could be a Indie game developer at that time.
05:21:03FromGitter<zacharycarter> I know - but I'm turrible at any artistic pursuit
05:21:12FromGitter<zacharycarter> I can watch a 3d modeling tutorial and replicate / change things a little
05:21:27FromGitter<zacharycarter> but when it comes to like - baking maps and actually making the model usable - I'm worthless
05:23:25FromGitter<bung87> ah,I once try developing a game found it need very much art resources.
05:25:19FromGitter<zacharycarter> well - I think you can get away with an idea and prototype art
05:25:40FromGitter<zacharycarter> so like - if you're making a 3d rts - generate the terrain etc and then for units use basic geometric shapes
05:25:48FromGitter<zacharycarter> and then get a concept down / working
05:26:10FromGitter<zacharycarter> once you have a game concept that's fun (you can work w/ a designer) then you can start attracting artists
05:26:20FromGitter<zacharycarter> artists will work on games that have good design behind them
05:26:32FromGitter<zacharycarter> sometimes for free if they like the idea enough and think it will take off
05:26:53FromGitter<zacharycarter> positive exposure = best thing ever for an indie dev
05:27:11FromGitter<bung87> yeah, that’s right,I just try to developing it see what problems I will meet during develping a game
05:27:20FromGitter<zacharycarter> there are lots ;)
05:27:35FromGitter<zacharycarter> but if you have any questions - or ever have an idea and want to collaborate with someone - let me know!
05:28:21FromGitter<zacharycarter> I have a lot of experience in that area - not experience with everything - but a lot of experience with building your own game library by combining other open source libraries to fill functionality needs
05:28:24FromGitter<bung87> like how to dymaticlly load map in a 2d or 3d game
05:28:36FromGitter<zacharycarter> I have a lot of experience with that kind of thing
05:28:41FromGitter<zacharycarter> both 2d and 3d
05:28:59FromGitter<zacharycarter> https://github.com/zacharycarter/mapgen
05:29:06FromGitter<zacharycarter> that was very early in my learning of proc gen days
05:29:29FromGitter<zacharycarter> not that I've done a ton since - but that was a fun project that ended up being a black hole
05:29:41FromGitter<zacharycarter> never ending trail of deepening complexity
05:33:00FromGitter<bung87> the map results look good.
05:33:41FromGitter<kayabaNerve> Is it possible to see if a ref was GC'd?
05:34:28FromGitter<kayabaNerve> I think the GC is clearing this ref early since accessing a string I can prove I set causes a segfault. I also can prove it's the same ref.
05:34:49FromGitter<kayabaNerve> (not saying it's me versus the community :P Saying I didn't make a typo; sorry if I was overly hostile there)
05:56:55FromGitter<Varriount> @kayabaNerve I believe there might be a GC debug symbol you can define.
05:58:36FromGitter<kayabaNerve> Which would do what specifically?
05:59:08FromGitter<kayabaNerve> So I parse it as X, which is a ref object of Y, cast it to X, and then cast it back to Y and pass it to an anonymous function stored in an object
05:59:25FromGitter<kayabaNerve> Said function can't access a member of Y, even though I know I set it and it's the same ref
05:59:44FromGitter<kayabaNerve> I tried GC_ref, declaring an var of Y and passing that var (not the cast)
06:00:14FromGitter<kayabaNerve> First was because I thought the GC was deleting it. Second was because what other option is there except manual memory management.
06:01:50FromGitter<kayabaNerve> And the member itself is a string
06:10:29*wildlander quit (Quit: Konversation terminated!)
06:21:25*narimiran joined #nim
06:23:51*radgeRayden quit (Ping timeout: 252 seconds)
06:42:11FromGitter<Varriount> @kayabaNerve Could you post the code?
06:45:29FromGitter<kayabaNerve> I don't have a MWE
06:48:00FromGitter<kayabaNerve> So I can link my work and specify files, but...
06:49:19FromGitter<Varriount> @kayabaNerve You say you parse it as X... Then cast it to X?
06:50:57FromGitter<kayabaNerve> Parse as Y ⏎ Return as X ⏎ Cast to Y ⏎ Pass as Y to anonymous function [https://gitter.im/nim-lang/Nim?at=5c53ec519221b9382df6f99e]
06:52:34FromGitter<Varriount> Y and X are reference types?
06:52:45FromGitter<kayabaNerve> X is a ref object ⏎ Y is a ref object of X
06:52:54*krux02 joined #nim
06:53:15FromGitter<kayabaNerve> I did modify X to be an object ⏎ Made Y an object of X ⏎ And then added Z as ref Y ⏎ ⏎ And then used Z where I previously used Y [https://gitter.im/nim-lang/Nim?at=5c53ecdbc2dba5382ec08aa8]
06:53:50FromGitter<kayabaNerve> And then I tried to created a new ref in the proc before I pass it, and memcpy the data out of the one that was returned, but then that segfaulted
06:54:27FromGitter<kayabaNerve> GDB commented on `genericAssignAux_mQY3YYo2RBkhPz09ajfLwpQ(void*, void*, TNimType*, bool) ()`
06:54:43FromGitter<kayabaNerve> And I'm trying to see what GDB thinks of my code without Z
06:55:30FromGitter<Varriount> I would really need to see the Nim code in question
06:56:24FromGitter<kayabaNerve> [Switching to Thread 0x7fffdf2c3700 (LWP 78380)] ⏎ copyString (src=0xf2f8f0045fc2a3ea) at /home/lukep/.choosenim/toolchains/nim-0.19.2/lib/system/sysstr.nim:100 ⏎ 100 if (src.reserved and seqShallowFlag) != 0:
06:56:38FromGitter<kayabaNerve> Yeah, I'll push and link it...
06:59:17FromGitter<kayabaNerve> @Varriount Mind if I PM it?
06:59:21FromGitter<kayabaNerve> And thanks for trying to help.
07:03:25FromGitter<Varriount> Sure
07:03:40avsejhi I saw new tag 0.19.4, but there is not tarball deployed. Is it like "pre-release" or something? or just someone forgot to upload the tarball on site?
07:04:54narimiranavsej: wait for couple of hours :)
07:07:48narimiranavsej: or use this if you're not patient: https://github.com/nim-lang/nightlies/releases/tag/version-0-19-2019-02-01-b6d96ca
07:19:03avsejthanks. I will check that, because previous release tarball was broken. and I had to skip that in Fedora
07:20:09narimiranavsej: please report if something is broken with 0.19.4
07:20:44avsejsure
07:24:35FromGitter<timotheecour> same thing with: `choosenim 0.19.4`: `Version 0.19.4 does not exist`
07:26:07avsejnarimiran, why there are several tarballs on the nightlies? what is the difference between -linux and -osx? which one is going to be published finally?
07:35:20FromGitter<mratsim> one is compiled for linux and the other for osx?
07:37:33*m712 quit (Ping timeout: 252 seconds)
07:41:03FromGitter<timotheecour> except no: ⏎ ⏎ > The Windows zips contain platform specific binaries whereas the the Linux and OSX archives contain pre-packaged source that can be used to build Nim on Linux, OSX and even Windows.
07:41:36*m712 joined #nim
07:41:58FromGitter<timotheecour> i hope in the future, it’ll build binaries instead of just src on linux and OSX (useful to speedup CI to avoid building over and over)
07:42:50FromGitter<timotheecour> should i re-open https://github.com/nim-lang/nightlies/issues/18 ?
08:00:00*gmpreussner quit (Quit: kthxbye)
08:04:53*gmpreussner joined #nim
08:05:14avseji've checked https://github.com/nim-lang/nightlies/releases/download/version-0-19-2019-02-01-b6d96ca/nim-0.19.4-linux.tar.xz and everything looks good
08:05:26avsejlooking forward tarball on the site
08:10:57*eizua joined #nim
08:12:01*eizua quit (Remote host closed the connection)
08:12:49*eizua joined #nim
08:13:27*eizua left #nim (#nim)
08:23:17FromGitter<mratsim> @Araq, if we want to write a image processing/tensor expression compiler that can generate LLVM IR in Nim, can we plug into Nim codegen with our own custom backend for a specific scope (macro)? reference https://github.com/mratsim/Arraymancer/issues/347, This is related to your recurrent GSoC idea for OpenCL code generator: https://github.com/nim-lang/Nim/wiki/GSoC-2016-Ideas#add-a-code-generator-for-opencl cc
08:23:17FromGitter... @timotheecour
08:27:10*kapil____ joined #nim
08:27:22FromGitter<timotheecour> @mratsim can an approach similar to this help ? https://github.com/fragcolor-xyz/nimline/blob/5ab07e505d60f105ad4d3f9930621a06ab31d403/nimline.nim#L126 which enables this: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c5402eaceb5a2264f762792]
08:29:34FromGitter<mratsim> Here I want to bypass the C++ generation and directly generate LLVM IR and join the happy ranks of: ⏎ Tensorflow XLA: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/compiler/xla/g3doc/overview.md ⏎ PyTorch Glow: https://github.com/pytorch/glow ⏎ Dask: https://dask.org ⏎ Gradient Halide: https://people.csail.mit.edu/tzumao/gradient_halide/ ... [https://git
08:29:35FromGitter... ter.im/nim-lang/Nim?at=5c54036e54f21a71a1dd84f1]
08:31:20FromGitter<mratsim> ideally in JIT modeas well.
08:34:13FromGitter<mratsim> SPIR-V IR is fine as well
08:40:08FromGitter<timotheecour> @mratsim How about: https://github.com/timotheecour/vitanim/blob/master/testcases/tests/t0127.nim#L7
08:42:56FromGitter<mratsim> I don't know much about compilers but I believe it's possible to pass the "what I want" and let the compiler optimize the "how to do it best" depending on the target platform but that may be a hacky first step, the main issue is that the C source file will have plenty of fluff (like OS related utilities)
08:43:51FromGitter<timotheecour> i just updated the snippet, it now works and outputs the llvm IR; is that what u want?
08:44:41FromGitter<timotheecour> generates: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c5406f8ca428b064537c275]
08:47:29Araqmratsim: I love the idea but currently it's a bit hard
08:47:52Araqwhat you really want is a "partial" codegen that deals with annotated procs only, I think
08:49:06Araqwith the compiler API you can add a new backend without touching the rest of the compiler, but it's lots of work and then needs to be able to compile everything
08:50:34*oculux quit (Quit: blah)
08:50:55*hwallann joined #nim
08:51:02FromGitter<mratsim> I want something like that in terms of high level API ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c540875ceb5a2264f76539b]
08:51:21FromGitter<mratsim> this is tensor addition
08:51:54*oculux joined #nim
08:52:11FromGitter<mratsim> inspired by my current forEach ⏎ ⏎ ```forEach x in a, y in b, z in c: ⏎ z = x + y``` [https://gitter.im/nim-lang/Nim?at=5c5408bb93fe7d5ac025612c]
08:57:28FromGitter<timotheecour> @mratsim r u expecting the body of `jit` to be arbitray nim code or is it a limited DSL that you control?
08:57:39FromGitter<mratsim> and higher level than my JIT with multiple backend target: https://github.com/numforge/laser/blob/master/examples/ex06_jit_hello_world.nim ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c540a03454aad4df7d8038d]
08:57:45FromGitter<mratsim> limited to a DSL
08:58:03FromGitter<timotheecour> if the former, this is similar to https://github.com/arnetheduck/nlvm project;
08:58:15FromGitter<mratsim> basically array/image processing DSL ;)
08:59:30FromGitter<mratsim> or if not a DSL, everything inside will be compute graphe nodes, that will be compiled/jitted at the end.
08:59:58Araqwrite an issue/RFC
09:00:11*Vladar joined #nim
09:00:12Araqbecause I don't really understand what you want, sorry
09:00:33FromGitter<mratsim> I want to reuse Nim codegen to output LLVM IR
09:00:45FromGitter<mratsim> the frontend part I'll take care of it.
09:00:53Araqbut it produces C(++) code
09:00:53FromGitter<timotheecour> is this related to halide https://github.com/mratsim/Arraymancer/issues/347?
09:00:58Araqand there is nlvm
09:00:59FromGitter<mratsim> yes it is
09:01:15FromGitter<timotheecour> then maybe we can follow up there?
09:01:23FromGitter<timotheecour> unless it’s more specific, idk
09:01:30FromGitter<mratsim> this is basically a followup on my latest post.
09:01:45FromGitter<mratsim> I'm exploring if Nim already as non-C/C++ codegen facilities
09:01:57FromGitter<mratsim> given the OpenCL codegen idea for GSoC
09:02:20Araqwell but openCL is close-ish to C and llvm IR ain't
09:02:26FromGitter<mratsim> but it seems like it's too rough so it's better if I implement my own codegen
09:02:33FromGitter<timotheecour> ya like i mentioned above (and so did araq) the most advanced project for that at this point would be nvlm https://github.com/arnetheduck/nlvm
09:04:13FromGitter<mratsim> I'm aware of it. I'll check how it does the codegen
09:05:13*floppydh joined #nim
09:05:36FromGitter<timotheecour> @araq can the nim VM generated IR be serialized ? if so than can a program compiled with regular nim, upon encountering a custom pragma, serialize its IR, passes it to nlvm and gets back LLVM IR >
09:08:03FromGitter<mratsim> I think we can just link to llvmruntime and write into a buffer
09:09:08FromGitter<timotheecour> > and write into a buffer ⏎ ⏎ but the underlying question is whether it’s serializable easily
09:10:46FromGitter<timotheecour> and maybe it is, i suppose it involves serializing a graph (representing nim compiler nodes) and avoiding serializing same node twice if 2 paths lead to same node
09:11:59Araqsounds like a terrible idea, the VM wasn't designed for native codegen
09:13:17FromGitter<mratsim> I probably only need something like that: https://github.com/numba/llvmlite
09:17:03FromGitter<mratsim> leaving for Brussels, see you this evening for those here.
09:19:58FromGitter<timotheecour> @mrastim how about this: ⏎ u write the nim frontend for your DSL: ⏎ ⏎ ```forEach x in a, y in b, z in c: ⏎ z = x + y``` ⏎ ... [https://gitter.im/nim-lang/Nim?at=5c540f3eceb5a2264f768366]
09:24:05*solitudesf quit (Quit: ZNC - https://znc.in)
09:25:05*solitudesf joined #nim
10:01:13*seni joined #nim
10:19:34FromGitter<arnetheduck> @mratsim isn't all the above simply `asm` or `emit` statements? more emit, with `nlvm`.. if you just want to emit IR, you don't want to write text either.. you want to use some llvm bindings and build them using the included builders
10:21:46FromGitter<timotheecour> i think he’s en route for brussels; but what he’s after is this: https://github.com/mratsim/Arraymancer/issues/347#issuecomment-459351890
10:23:30FromGitter<timotheecour> @arnetheduck your approach requires compiling with `nlvm` right? haven’t tested in a while so idk how mature it is?
10:40:53*JustASlacker joined #nim
10:43:05FromGitter<arnetheduck> `emit`/`asm` would need `nlvm` - it doesn't support either right now. if you just want to spit out llvm ir (even at compile time), you could do that with a wrapper as long as ffi is enabled
10:54:09FromGitter<timotheecour> sorry it’s super later here and my brain is fried; ⏎ ⏎ > emit/asm would need nlvm - it doesn't support either right now ⏎ ⏎ 1) u’re saying llvm doesn’t support `emit/asm` right now right? so can’t be used for that ? ... [https://gitter.im/nim-lang/Nim?at=5c54255178e1ed4103c400e4]
10:59:14Araqhttps://forum.nim-lang.org/t/4610 can somebody help him?
11:02:54FromGitter<timotheecour> Done => https://forum.nim-lang.org/t/4610#28858 `import segfaults` shd help
11:09:37Araqlol I don't think so, but thank you
11:16:21Zevvwhy is 'segfaults' nowhere in the docs?
11:17:32FromGitter<timotheecour> PR welcome
11:17:51FromGitter<timotheecour> it’s very useful, when it’s time to debug some segmentation faults
11:17:59Zevvsorry, I ment: why does it not end up in the docs? it has a ## doc block?
11:18:48FromGitter<timotheecour> for same reason testament doesn’t run tests on everything: it doesn’t run docs on ‘**/*'
11:18:55Zevvah ok
11:18:56FromGitter<timotheecour> ```**/*```
11:19:27FromGitter<timotheecour> that’s the issue: https://github.com/nim-lang/Nim/issues/10276
11:24:42*ng0 joined #nim
11:28:45FromGitter<GULPF> segfaults is in lib/pure, so that's not the issue
11:28:59FromGitter<GULPF> the docs are generated, they are just not linked anywhere I guess: https://nim-lang.org/docs/segfaults.html
11:29:58*Perkol joined #nim
11:31:38*dom96_w joined #nim
11:31:44FromGitter<timotheecour> inded; `segfaults` is not in https://nim-lang.github.io/Nim/theindex.html and searching for `segfaults` returns nothing in searchbar
11:32:05FromGitter<timotheecour> => we should allow searching for modules in search bar ⏎ => index.html should index modules
11:34:19*nsf quit (Quit: WeeChat 2.3)
11:36:05FromGitter<timotheecour> weird, other modules seem listed; just some (like segfaults) are not
11:55:57*kapil____ quit (Quit: Connection closed for inactivity)
11:56:53dom96_wFOSDEM time yay
11:57:01*dom96_w quit (Quit: Textual IRC Client: www.textualapp.com)
12:14:48federico3any tip on implementing a callback from a dynamic library?
12:19:24Araqget the calling convention right
12:20:03federico3any example?
12:22:40*dddddd joined #nim
12:24:30Araqproc calledFromC(arg1: cstring; arg2: cint) {.cdecl.}
12:24:47Araqsorry if that was obvious and unhelpful
12:31:59Zevvfederico3: what's your specific problem?
12:32:47ZevvThe callback should be fairly trivial, but take care if the lib starts calling you from a different thread, for example
12:39:43*stefanos82 joined #nim
12:42:17federico3I have to catch my flight, thanks for the help
13:15:14*kapil____ joined #nim
13:20:28narimiranAraq: check your local repo, you might be one commit ahead of master
13:20:58narimiranheh, out of context. wrong channel :)
13:29:17*Sentreen quit (Quit: WeeChat 1.6)
13:31:59*vlad1777d joined #nim
13:36:04*Cthalupa quit (Ping timeout: 246 seconds)
13:37:26*Snircle joined #nim
13:37:29*Cthalupa joined #nim
13:40:41*Sentreen joined #nim
13:41:21*Perkol quit (Quit: Leaving)
13:41:53*Cthalupa quit (Ping timeout: 246 seconds)
13:43:28*Cthalupa joined #nim
13:44:49*abm joined #nim
13:46:13shashlickNarimiran - I'll be working on nightlies today
13:46:25shashlickWhen is 0.19.4 going to be released?
13:46:32shashlickI want to add Linux binaries
13:46:44narimiranshashlick: there's more work to be done?! we planned to use the stuff that's already there
13:46:56narimiranwhen? this afternoon :D
13:47:04narimiranbut we can wait, i guess :)
13:47:11shashlickWell I did a poc with holy build box
13:47:34shashlickSo we can have Linux binary xz with everything pre built
13:47:35narimiranbinaries would be nice, i guess
13:47:54shashlickThing is then we will need to post separate xz for osx
13:51:51FromGitter<arnetheduck> @timotheecour `nlvm` has nightly releases btw: https://github.com/arnetheduck/nlvm/releases
13:52:59FromGitter<arnetheduck> and `emit` support would have to be added (I'd probably implement it a different way than emitting text though)
13:53:27*stefanos82 quit (Remote host closed the connection)
14:01:10shashlick@arnetheduck - I can take up enabling llvm fully in nimterop, will help make C support more complete but will you be able to start testing and using it?
14:05:37FromGitter<bung87> user define type name will conflict with std lib ?I got`/Users/bung/Nim/lib/pure/net.nim(138, 15) Error: ambiguous identifier: 'Protocol' --use protocol.Protocol or nativesockets.Protocol`
14:10:36*hwallann quit (Remote host closed the connection)
14:13:41FromDiscord_<juan_carlos> Hi.
14:15:39*floppydh quit (Quit: WeeChat 2.3)
14:16:05*radgeRayden joined #nim
14:17:54FromGitter<mratsim> @arnetheduck @timotheecour yes I was thinking on directly linking to LLVM runtime and use it’s code builder facilities
14:18:37FromGitter<mratsim> I found out that LLVM has something called “bitcode” which is a binary representation of its IR so I wouldn’t have to go through a text file.
14:19:29FromGitter<mratsim> This would also be a good match to the JIT Assembler I started, except that I would use it for LLVM IR/bitcode instead of x86_64 assembly
14:22:34FromGitter<mratsim> Also Nvidia describes the LLVM IR that is valid for GPU generation: https://docs.nvidia.com/cuda/nvvm-ir-spec/index.html hopefully this is also valid for OpenCL and AMDGPU target.
14:48:28*Cthalupa quit (Ping timeout: 250 seconds)
14:49:59ZevvNim vertex shaders :)
14:51:04*Cthalupa joined #nim
14:55:55FromGitter<mratsim> I don’t know anything about Graphics programming so can’t say :P
14:56:23FromGitter<mratsim> if anything I’ll subvert them for machine learning (especially on Mac and mobile devices)
15:00:27FromDiscord_<demotomohiro> You can not write vertex shaders with CUDA
15:01:47ZevvI know, I know
15:01:57livcdis playground broken ? I cant seem to run this: https://glot.io/snippets/f934240p60
15:02:03Zevvbroken for some weeks
15:02:07Zevvjust as nimbot
15:03:04FromGitter<mratsim> @demotomohiro I think Zevv was talking about jitting LLVM IR into vertex shaders
15:03:32livcdah that's a shame
15:03:59ZevvNah, not really. not much to JIT, as shaders are typically text which gets compiled by the GPU drivers
15:04:13ZevvI was just trolling about
15:04:37Zevvlivcd: Zah was working on play/bot some time ago, not sure what the status is
15:04:54FromDiscord_<juan_carlos> https://tio.run/#nim
15:05:40FromDiscord_<demotomohiro> LLVM IR can be used to write vertex shaders?
15:06:01FromDiscord_<demotomohiro> I know OpenGL supports SPIR-V
15:06:08ZevvTio also doesn't like "import posix; while true: discard fork()" :)
15:06:28narimiranthat's why we can't have nice things ;)
15:08:20FromDiscord_<juan_carlos> Dont break the Nim 😥
15:08:34Zevv*someone* has to break it
15:08:45*Vladar quit (Read error: Connection reset by peer)
15:08:48ZevvOh it has proper ulimits "Failed to copy files from the runtime temporary directory
15:08:49ZevvUnable to fork: Resource temporarily unavailable"
15:08:50Zevvthat's good
15:09:12FromDiscord_<juan_carlos> Fun b'cause Tio Nim is literally "Uncle Nim" on Spanish.
15:09:20livcdjuan_carlos: tanks but cant turn on threads
15:11:02FromGitter<mratsim> @demotomohiro, probably: https://github.com/LunarG/LunarGLASS
15:11:04*Vladar joined #nim
15:17:09FromDiscord_<demotomohiro> @mratsim thank you! It seems LLVM IR can be converted to SPIR-V.
15:17:36FromGitter<mratsim> AFAIK it’s still work in progress for an official SPIR-V backend
15:18:01Araqhttps://www.youtube.com/watch?v=sEQnmInxmfg
15:18:08Araqas promised
15:18:38FromDiscord_<juan_carlos> 👀🍿
15:18:41Zevv\o/ !
15:19:22*JustASlacker quit (Remote host closed the connection)
15:19:33*leorize quit (Ping timeout: 252 seconds)
15:19:40Araqand this time I cut out krux02's ramblings ;-)
15:19:54*leorize joined #nim
15:20:41FromGitter<mratsim> look who’s talking :P
15:24:33narimirannim dev blog, directors cut :D
15:25:07Araqlol
15:26:22FromGitter<mratsim> “Hello World works” yeaaah
15:28:41FromGitter<mratsim> was the parsing bug the reason I had to do: ⏎ ⏎ ```code paste, see link``` ⏎ ⏎ ? [https://gitter.im/nim-lang/Nim?at=5c5465a954f21a71a1e020fc]
15:36:21Araqno. :-)
15:36:41Araqthat's because the command *expression* syntax only supports one argument
15:36:52Araqin order to not confuse anybody :P
15:37:10FromGitter<bung87> it fix proc_call [1,2,3] with array
15:37:45FromGitter<bung87> otherwise need proc_call([1,2,3])
15:42:53livcdcool podcast
15:44:28Araqthank you
16:04:25FromGitter<riddl_gitlab> @Araq those changes you are talking about on podcast, e.g. fixed syntax for function calls with tuples are going to be available with 0.19.4 ? or is it only on devel branch for now?
16:06:50Araq0.19.4 was released today but we're still updating the website :P
16:07:21Araqand it's a bugfix release, technically the parser thing was a grammar extension so it wasn't backported
16:07:44Araqthe 0.19.x series is about boring conservative stuff that really cannot make anything worse
16:08:00Araqall the new exiting things are for 0.20
16:08:20shashlickok so I missed the boat on adding linux binaries - anyway makes more sense on the 0.20 release
16:08:32FromGitter<riddl_gitlab> sure, thanks
16:08:38FromDiscord_<demotomohiro> Are there source code of nim 2 glsl compiler macro?
16:08:41shashlickso Araq, right now both linux and OSX get the same link for download
16:08:49AraqI know
16:08:56Araqalready uploaded, shashlick
16:09:04FromDiscord_<demotomohiro> I cannot find about nim 2 glsl with google
16:09:10shashlickif we create a linux specific binary package, are you open to a separate OSX source package
16:09:29Araqyep
16:09:33*abm quit (Remote host closed the connection)
16:09:43Araqbut it also needs a tiny website redesign
16:09:50Araqright now we have Unix vs Windows
16:09:54*abm joined #nim
16:10:19Araqbah
16:10:33Araqhttps://nim-lang.org/install.html still says we have no nightly builds :-(
16:10:37Araqbut we do...
16:10:40FromGitter<riddl_gitlab> `proxychains choosenim update stable` ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5c546f807b68f9410254f2ea]
16:10:57FromGitter<riddl_gitlab> something is not working well
16:11:51FromGitter<riddl_gitlab> `wget http://nim-lang.org/download/nim-0.194.tar.xz` ⏎ ```2019-02-01 17:11:23 BŁĄD 404: Not Found.```
16:12:19narimiran@riddl_gitlab choosenim is not yet updated, AFAIK
16:12:24FromGitter<riddl_gitlab> i doubt ts fault of my corp's proxy
16:12:28FromGitter<riddl_gitlab> oh, ok
16:12:29Araqhttps://nim-lang.org/download/nim-0.19.4.tar.xz
16:12:34Araqit is missing a dot for some reason
16:13:03narimiraneither the dot is missing, or there is one dot too much
16:13:15narimiranshould be -0194 or -0.19.4
16:13:47FromGitter<riddl_gitlab> yeah, that link is working. For some reason choosenim is looking for -0.194...
16:14:18shashlickwhat OS @riddl_gitlab
16:14:32shashlickand what version of choosenim
16:14:44Araqah thanks, fixed
16:14:45*abm quit (Ping timeout: 268 seconds)
16:14:46FromGitter<riddl_gitlab> choosenim v0.3.2 (2018-02-28 14:12:37) [linux/amd64]
16:14:48Araqcan you retry please?
16:15:24FromGitter<riddl_gitlab> Couldn't download file from http://nim-lang.org/download/nim-0.1.4.tar.xz.
16:15:25FromGitter<riddl_gitlab> ;D
16:15:28FromGitter<riddl_gitlab> 99999
16:15:32FromGitter<riddl_gitlab> missing now ;)
16:15:49Araqcan you retry please?
16:16:11FromGitter<riddl_gitlab> Works :)
16:16:13AraqI'm too stupid to type in the version number correctly :D
16:16:43FromGitter<riddl_gitlab> you need some automation for that :D
16:18:11shashlickdoes anyone know if OSX binaries work across versions or need some special work like on Linux
16:19:17Araqshashlick, no idea. I would guess they work rather well without any shenanigans
16:19:47shashlickok so nightlies already creates separate files so we simply need web changes
16:20:06shashlicki'll enable saving binaries in koch xz and we can test them over this cycle and see how well they work
16:20:24shashlicklinux will have to run a docker container
16:20:35shashlickbut it is quite simple
16:20:43FromGitter<riddl_gitlab> i need to read how to build a flatpak nim app..
16:21:03FromGitter<dom96> You'd think writing 0.19.4 in a text file wouldn't be that hard :p
16:21:36narimiranburn from Brussels?
16:22:01FromGitter<riddl_gitlab> haha, give him a break :D typos happens
16:22:03FromGitter<arnetheduck> fwiw, `nlvm` builds a single-binary file that will work on all linuxes and includes the std lib and all, using travis
16:22:06FromGitter<dom96> I was hoping I could work on this in the train but the connectivity is incredibly bad
16:24:17shashlick@arnetheduck - how are you bundling everything together including stdlib
16:24:28FromGitter<arnetheduck> appimage
16:25:12FromGitter<arnetheduck> https://github.com/arnetheduck/nlvm/releases, https://github.com/arnetheduck/nlvm/blob/master/.travis.yml
16:25:22shashlickis it like pyinstaller / extracting everything into temp every run type scenario?
16:26:56*noonien joined #nim
16:27:13FromGitter<arnetheduck> squashfs, mounted using a loop device
16:29:05*Trustable joined #nim
16:29:07shashlicknice - using holy build box is also easy, http://ix.io/1zIU is how you run it and http://ix.io/1zIS is how i'm building the package
16:29:27shashlickbut ya, typical xz file rather than single binary
16:30:07shashlicki want to create a portable nim distro - nim.exe, minimal stdlib and tcc
16:30:10*vlad1777d quit (Ping timeout: 268 seconds)
16:32:07shashlickbut looking at windows for starters
16:32:19FromGitter<dom96> Are you planning to distribute Nim nightlies with a dependency on docker?
16:32:49shashlickno, it uses docker to load an old distro
16:32:52radgeRaydensorry if this is too simple but is there a tool for updating on windows or am I supposed to do it manually? I saw you guys talking about choosenim but I don't seem to have this.
16:33:05shashlickand builds within that so the resulting binaries are compatible with many new distros
16:33:12shashlickdocker only for compile
16:35:17narimiranradgeRayden: you can do it manually. if i remember correctly, it is "2-click process", nothing complicated
16:35:36radgeRaydenso just open the installer and it should be fine, thanks
16:49:58*darithorn joined #nim
16:57:01leorizeshashlick: do you link it with glibc?
16:57:38shashlickI believe it does
16:57:59shashlickhttps://github.com/phusion/holy-build-box#which-linux-distributions-do-binaries-compiled-with-holy-build-box-support
16:58:54Calinoushashlick: tcc is probably a bad choice if you want high performance though
16:59:04shashlickmany ways to solve this problem but this method took me an hour to pull off
16:59:37Calinoualso, you can create a .tar.xz with a single command, use `tar cfJ archive.tar.xz directory/`
16:59:45shashlickCalinou - I just want a mini nim to build a plugin system
16:59:55Calinou(level 9 is pretty much unnecessary, and it requires more RAM to be extracted too)
17:00:08Calinouit's not like level 6 is too fast already :P
17:00:12shashlickya, I'll be extending koch xz to productize this
17:00:21shashlickjust leveraged what was in koch xz for that POC
17:00:42leorizeI don't really think you should package a C compiler with
17:01:05shashlickthat's not for the nightly or official releases, just for a POC text editor i'm working on
17:01:20shashlickwant to build a plugin system
17:01:48leorizea nimscript based plugin system?
17:01:59leorizeor a proper C interface?
17:02:04shashlickthat's where I started, but --app:lib works so nicely
17:02:36shashlickanyway, will see - better to ship plugins as binaries
17:03:16leorizebinary plugins aren't the best when it comes to cross-platform
17:04:06shashlicktrue but i'll have to provide a build of the editor anyway
17:13:47*leorize quit (Quit: WeeChat 2.3)
17:34:24*tobbez quit (Quit: Lost terminal)
17:35:38Zevvfosdemmers: happy beerfest!
17:35:58*kapil____ quit (Quit: Connection closed for inactivity)
17:44:03*leorize joined #nim
17:45:35*nsf joined #nim
17:46:01*tobbez joined #nim
17:46:23FromDiscord_<juan_carlos> Holy Box is not like the similar to AppImage ?.
17:50:33shashlickAppImage has squashfs and bundles everything into one file
17:50:43shashlickholy build box just simplifies building with an old glibc
17:52:01FromDiscord_<juan_carlos> 🆗
17:58:37FromGitter<zacharycarter> https://github.com/idea4good/GuiLite
18:00:38FromGitter<zacharycarter> have our prayers finally been answered (kind of anyway)?
18:01:27Zevvwhat prayers?
18:01:41FromGitter<zacharycarter> cross platform GUI solution that doesn't suck
18:01:54narimiranZevv: prayers for yet another GUI lib :D
18:01:56FromGitter<zacharycarter> and offers retained mode
18:02:16FromGitter<zacharycarter> not yet another - one that actually does what it needs to well
18:02:30FromGitter<zacharycarter> and this seems to show promise - but will only know after its used by more people I suppose
18:03:04ZevvI was already ready to type "nuklear and imgui" :)
18:03:45FromGitter<zacharycarter> ah - yeah but those are immediate mode
18:03:54Zevvyes indeed
18:03:54FromGitter<zacharycarter> we need something like Qt
18:03:58FromGitter<zacharycarter> but with less Qt :P
18:04:12Zevvis that something like wxwidgets, with less wx?
18:04:27FromGitter<zacharycarter> heh - wxWidgets is okay I guess but it's a large dependency
18:04:28FromGitter<zacharycarter> just like Qt is
18:04:37Zevvalways the problem
18:04:49FromGitter<zacharycarter> ideally we have something that is small, and doesn't concern itself with all of the bells and whistles that Qt and wxWidgets do
18:05:03shashlickno example?
18:05:06FromGitter<zacharycarter> for Nim GUI solutions - I still think nigui is the best
18:05:15FromGitter<zacharycarter> yeah - that's what was bugging me - it's hard to make sense of the demos
18:05:30FromGitter<zacharycarter> he says like - this library doesn't actually build the UI controls - you have to do that yourself
18:05:32FromGitter<zacharycarter> so I'm a bit puzzled
18:06:00FromGitter<zacharycarter> `GuiLite just a framework, it could not create UI by itself. In order to show how to build UI App with it, We provide some samples for your reference.`
18:06:08FromGitter<zacharycarter> https://github.com/idea4good/GuiLiteSamples
18:07:06ZevvHow is this different from all the others?
18:07:15FromGitter<Vindaar> @Araq in case you're here. I'm looking at https://github.com/nim-lang/Nim/issues/10478. I'm confused by your comment. The macro is evaluated but during `semAfterMacroCall` the call to `semExpr` fails, because `nkExprEqExpr` isn't handled at all in that proc. Are you saying we should never end up in `semExpr` in the first place?
18:07:34FromGitter<zacharycarter> seems much lighter weight
18:07:44Zevvok, true, if its really 5kloc
18:07:46FromGitter<zacharycarter> also - it has a C API - so it should be easy to bind to
18:07:52FromGitter<zacharycarter> that's not true with Qt
18:07:55shashlicki'm just wondering what header has teh API
18:07:58Zevvhttps://github.com/andlabs/libui
18:08:06FromGitter<zacharycarter> libui is very buggy from what I've heard
18:08:08FromGitter<zacharycarter> and unfinished
18:08:08shashlickcause they just extern in the user app
18:08:11FromGitter<Varriount> TheLemonMan: Have you seen https://probablydance.com/2018/06/16/fibonacci-hashing-the-optimization-that-the-world-forgot-or-a-better-alternative-to-integer-modulo/ ?
18:08:43FromGitter<zacharycarter> this may very well not be any better - but who knows - it's another potential option to check out and play around with
18:09:08Zevvtrue
18:09:59FromGitter<Varriount> libui had such promise
18:10:43ZevvI made the Lua binding once, but abandoned it. It was indeed not finished and buggy
18:10:49Zevvbut still being developed now, it seems
18:15:17*TheLemonMan joined #nim
18:16:53TheLemonManVarriount, yeah but the savings are negligible until your table gets huuge
18:17:56TheLemonManI wanted something nice for small & medium-sized dicts
18:19:33TheLemonMangotta experiment with robin-hood hashing for the indices array tho but I expect it to be slower than the current probing scheme
18:24:47TheLemonManVindaar, yeah the problem is not-that-easy to tackle since the macro is expanded during the parameter evaluation pass
18:26:12FromGitter<Vindaar> @TheLemonMan Ah, thanks. Any rough idea how to proceed?
18:27:48TheLemonMannope
18:27:55FromGitter<Vindaar> Hehe, ok :)
18:35:55shashlicknarimiran: windows nightly is failing due to windows path differences
18:36:03narimiranshashlick: gaaah
18:36:33narimirani'll put that example behing `when defined(posix)`
18:36:45narimiranit is `/../`, right?
18:36:56shashlickya https://github.com/nim-lang/Nim/blob/devel/lib/pure/os.nim#L428
18:37:09shashlickand you get a/b\d/e
18:37:13shashlickjust because of that one \
18:37:20shashlickeven the other one might break the same way
18:37:28shashlickyou could just replace \ with /
18:38:06shashlickyou get a\d/e for second one
18:38:59narimiranshashlick: i pushed the quickfix
18:39:50narimiraneverything else in `os.nim` is ok when you do `nim doc`?
18:40:17AraqTheLemonMan, how about implementing BTree.del ?
18:40:38AraqI want BTrees in the stdlib, no hashing, no cry
18:40:42shashlickthere's more
18:40:59shashlickline 1256
18:41:04narimiranargh, give me the whole list
18:41:05Araqdeterministic iteration order, no security attack vector
18:42:14shashlicknarimiran: looks like normalizePath doesn't work
18:42:19shashlicka does not change on windows
18:42:20narimiranshashlick: and few lines below also, probably `normalizeD`
18:42:35shashlickif I echo a after normalizePath(), echo a => a///b//..//c///d
18:42:58narimirananything more after `normalize(d)`?
18:43:23narimirani'm still surprised and disappointed that appveyor didn't catch that stuff
18:43:24shashlick1293 also fails
18:43:41narimiranthat's `normalized`, right?
18:43:42shashlickthat's it
18:43:47shashlickyes
18:44:18narimiranwhen you remove those three (/../, normalize, normalized), everything else is ok?
18:44:32shashlickyes
18:44:36narimiranok, pushing
18:44:46shashlickbut normalize isn't working at all, so that needs to be fixed
18:51:52TheLemonManAraq, have you seen my port of Python's dict implementation?
18:51:58Araqnope
18:52:11TheLemonManhttps://github.com/LemonBoy/compactdict here you go
18:52:22TheLemonManmuch simpler than btrees
18:52:38AraqI don't agree.
18:53:02Araqyou never know the hashing's attack vectors
18:53:08Araqand btrees don't have any
18:54:45TheLemonManyeah, that's true, but with a good-enough™ hash plus some clever shuffling you can go pretty far
19:00:05Araqany benchmark numbers?
19:00:17Araqis it better than Nim's stdlib?
19:01:30Araqtemplate nextTry(h, maxHash: Hash): untyped = ((5*h) + 1) and maxHash # we used to do this but it's pretty bad because of the cache behaviour
19:02:03TheLemonMangotta implement rehashing to a smaller size first and I'll benchmark it
19:02:48narimiranTheLemonMan: did you forget `.toDict()` in your readme example?
19:02:51TheLemonManyeah but this time the array is a tad smaller so it's more likely to stay in cache
19:03:22TheLemonMannarimiran, yep, too bad markdown doesn't support the :test: directive :P
19:14:17*abm joined #nim
19:23:44TheLemonManplotly api is nightmare fuel
19:25:46AraqI hate the Element vs Node distinction in the DOM, it's crap
19:26:40TheLemonMansee, we're _never_ happy!
19:39:52*natrys joined #nim
19:45:53*d10n-work joined #nim
19:46:07AraqTheLemonMan, btw your PRs are a bit on hold until I've implemented nimble package checking logic to detect more regressions
19:47:05TheLemonManno worries, better be safe than sorry
19:56:41FromDiscord_<exelotl> woooo markdown features in docgen!
19:56:52FromDiscord_<exelotl> that's awesome <3
20:00:37*abm quit (Quit: Leaving)
20:11:20FromGitter<basp> Agreed. I think the whole language and toolkit is awesome to work with.
20:15:27Araqthanks.
20:15:38Araqwe also have a new milestone btw: https://github.com/nim-lang/Nim/milestone/6
20:17:59rayman22201👍 to the new milestone
20:22:34*TheLemonMan quit (Ping timeout: 250 seconds)
20:26:09FromGitter<basp> excited for release candidate, thank *you*!
20:30:59*wildlander joined #nim
20:31:04*wildlander quit (Changing host)
20:31:04*wildlander joined #nim
20:42:24Zevvis it feasible to backport the doc generator with linkable procs to 0.19.4 so that the current docs get a bit more friendly while the big revamp is going on behind the scenes?
20:44:24narimiranre markdown: it is still rough around the edges and some stuff don't work (but even some stuff don't work in rst :P)
20:44:34narimiranZevv: glad you like that feature
20:45:03narimirani took me 2 days of explaining what i want to have implemented :D
20:45:18ZevvI never realized it wasn't there, but now that it is I realize how much it helps
20:45:24ZevvWhy is it hard to explain?
20:45:33narimiranbecause i'm retarded :D
20:45:46Zevvcongrats, that makes two of us
20:46:04narimiranbtw, docs CSS revamp will make much easier to spot these "hidden" links
20:46:39narimirancurrent links work by the word of mouth: "hey, did you know that we now have links there and there?"
20:46:54Zevveuh, they do stand out in blue?
20:47:00narimirannope
20:47:05Zevvoh the other ones, true :)
20:47:14shashlick@kaushalmodi - travis caching is not working on windows
20:47:15ZevvI just opened a page to see that they're not blue :)
20:47:32narimiranthat would be too much, but they'll have a nice mild underline
20:47:51narimiranso you know they are links, but no in-your-face screaming
20:48:24Zevvnice.
20:48:41ZevvAre you open for input still?
20:49:04narimiranalways ;)
20:49:10narimiranwhat do you have in mind?
20:49:20Zevvnot sure if it makes sense:
20:49:29Zevvthe 'see also' sections now say "this proc" and "that proc"
20:49:32FromGitter<kaushalmodi> shashlick: I'd resort to opening an issue on the travis ci repo
20:49:34xaceany nim lib that allows you to bind global hotkeys in a X11 environment in linux?
20:49:40Zevvthat's quite verbose, and proc is the norm
20:49:42FromGitter<kaushalmodi> they usually respond in a day or two
20:50:10FromDiscord_<juan_carlos> Why instead of maintaining a CSS, dont use some CSS Framework thats minimalistic and dont need JavaScript?, like Spectre (9kb) or Lit (300Bytes).
20:50:12ZevvWould it make sense to make it this() and that(), and maybe only note the exeption for templates and macros
20:50:21Zevvoh of course there is also iterators, funcs, etc, aw
20:50:33Zevvno nevermind
20:50:34FromGitter<kaushalmodi> shashlick: I don't use Windows so it's difficult to guess what's wrong in the current config
20:50:46Zevvjust thinking out loud
20:50:50narimiranah, i started with that `... proc` stuff, is seemed suitable, but i get your point
20:50:56FromGitter<kaushalmodi> may be `[[ -f FILE ]]` where FILE has paths with forward slashes doesn't work?
20:51:03narimiranbut that is not CSS, that's just lots of manual deleting :D
20:51:35ZevvAre these verbatim in the docs?
20:51:47narimiran@juan_carlos because that would be much more work for little benefit? :/
20:53:00rayman22201@xace, there are x11 bindings, but no docs, and I don't know enough about it to help. Maybe this will help anyway? https://github.com/PMunch/x11
20:54:23Zevvglobal hotkey hooking requires record X11 extension
20:54:25narimiranhmmm, now i noticed that devel docs, when you click on `source` takes you to **master** branch, not devel :(
20:56:46Zevvxace: You need only a handful of X11 calls for key hooking, if that helps you I can see if I can make a quick wrapper for you?
21:00:13*TheLemonMan joined #nim
21:00:28*Tyresc joined #nim
21:03:21*zachk joined #nim
21:15:30shashlick@kaushalmodi - that's not the issue
21:15:48shashlickif you see the beginning where cache gets restored, there's nothing getting restored, which means nothing gets backed up
21:17:11FromDiscord_<juan_carlos> How can be the best way to generate a random string of 100 ascii letters?.
21:20:05shashlickthe syntax to cache is obviously correct since it works on Linux and OSX
21:20:10shashlickbut gets left out on Windows
21:20:22shashlickconsidering windows is new for travis, I'm wondering if it doesn't work yet
21:20:34shashlickhaven't found any known issues though
21:30:41*Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
21:35:14FromGitter<Varriount> Araq: What is the alternative to multi-methods?
21:37:59Zevvjuan_carlos: something like var s: string; for i in 1..100: s.add random(33..127).char might do?
21:38:59*nsf quit (Quit: WeeChat 2.3)
21:41:10FromDiscord_<juan_carlos> I used:
21:41:15FromDiscord_<juan_carlos> `repeat($rand(10_00_00_00_00_00_00_00_00.int..int.high), 5)`
21:41:27FromDiscord_<juan_carlos> Dont need to be crypto strong.
21:42:05*TheLemonMan quit (Quit: "It's now safe to turn off your computer.")
21:42:21ZevvI guess that also works :)
21:55:55Araqvarriount: single dispatch methods
21:59:27FromGitter<Varriount> Araq: How are those better?
22:00:12Araqthey are a bit easier to understand and we can later implement them via more traditional vtables
22:00:46Araqand we should also get rid of the strange .base annotation then
22:01:12Araqit's not a big deal, people claimed it won't break any code besides the tests
22:29:28FromGitter<iffy> Before I spend a bunch of time debugging: is cross-compiling a static lib from macOS for Windows supported?
22:31:15Araqunlikely
22:31:28Araqyou would need a clang/gcc that can do it
22:32:11*abeaumont quit (Ping timeout: 250 seconds)
22:34:16FromGitter<iffy> and mingw doesn't?
22:34:35FromGitter<iffy> nim happily makes the static lib, btw -- it just seems to have trouble be used when I get to Windows
22:42:36shashlickhow does winrelease build tools? i don't see it called anywhere
22:44:51shashlickok it calls koch zip which does
22:49:24FromDiscord_<akavel> Hi! I'm a total newbie. I want to read and write some binary data from stdin/to stdout, in big endian. How would I best approach this? I want to parse/emit a particular protocol.
22:50:20FromDiscord_<akavel> (It's mostly big endian uint64 numbers, interspersed with raw strings/byte spans)
22:52:05FromDiscord_<akavel> Also, I would like to be able to write tests for it, so it should be possible to substitute some mock input in place of the stdin, and verify that output is correct (matching some session transcript)
22:52:11*craigger quit (Quit: bye)
22:53:09FromDiscord_<akavel> (I can write my own mocks, I am just not sure how to structure the thing to make sure the stdin/stdout can be easily replaced with my mock object/type/funcs/...)
22:53:38FromDiscord_<akavel> Could you please help?
22:55:14shashlickstdin/out are just files so you could stub with just files
22:56:00shashlickalso, simple readChars/readBytes or the like can be used to read and write
22:56:28shashlickthere's also an endian lib that can help
22:57:07*narimiran quit (Ping timeout: 240 seconds)
22:59:41FromDiscord_<akavel> Hm; could you possibly point me to some funcs in Nim stdlib docs that could get me started?
22:59:49FromDiscord_<akavel> I must admit I have trouble navigating them :/
23:00:45FromDiscord_<akavel> So, as much as I maybe grasp your general thoughts, I still don't have an idea how to approach this in Nim 😦
23:01:47FromGitter<kaushalmodi> akavel: https://nim-lang.github.io/Nim/theindex.html and Ctrl+F in browser to the rescue :)
23:02:19FromGitter<kaushalmodi> agreed that the documentation of https://nim-lang.github.io/Nim/endians.html an use some help
23:02:25FromDiscord_<akavel> Ok, thanks, I'll try it! 😃
23:02:36FromGitter<kaushalmodi> s/an/can
23:03:22FromGitter<kaushalmodi> shashlick: sorry, couldn't spend any time at looking at the windows failure; too much work
23:03:27FromDiscord_<akavel> Also, if I were to use Streams, can I somehow pass stdin & stdout to them?
23:04:58FromDiscord_<akavel> Hmmm, ok, I think I found it myself:
23:05:11FromDiscord_<akavel> thanks to the index, I found that 'stdin' is a 'File',
23:05:51FromDiscord_<akavel> and the 'streams' library has a 'newFileStream()' variant taking a 'File' - so, I suppose they should work together 😉
23:06:00FromDiscord_<akavel> thank you guys!!!
23:06:09*craigger joined #nim
23:06:20FromGitter<kaushalmodi> you can look at https://github.com/nim-lang/Nim/blob/devel/lib/pure/unittest.nim
23:06:25*craigger quit (Client Quit)
23:06:26FromGitter<kaushalmodi> it is using streams
23:06:34FromGitter<kaushalmodi> I haven't studied that code
23:06:43*krux02 quit (Remote host closed the connection)
23:06:44FromGitter<kaushalmodi> but I know it has a bug where it does not send errors to stderr
23:07:03FromGitter<kaushalmodi> may be if you figure out streams and std*, you can patch it :P
23:07:05FromGitter<kaushalmodi> no pressure
23:07:07FromGitter<kaushalmodi> :)
23:07:37FromGitter<kaushalmodi> ref: https://github.com/nim-lang/Nim/blob/devel/lib/pure/unittest.nim
23:07:49FromGitter<kaushalmodi> wrong ref
23:07:54FromGitter<kaushalmodi> right ref: https://github.com/nim-lang/Nim/issues/7913
23:09:28*craigger joined #nim
23:09:34FromGitter<Vindaar> > <TheLemonMan> plotly api is nightmare fuel ⏎ Are you talking about nim-plotly or plotly.js? If the former, sorry. That's partly my fault. I'm all open ears to improve it though! If the latter, well yeah. It's not great...
23:09:44FromGitter<Vindaar> ugh, gitter..
23:09:45*craigger quit (Client Quit)
23:09:58shashlick@kaushalmodi - i've been doing some tests
23:10:07shashlickhttps://github.com/genotrance/nightlies/pull/1
23:10:55*natrys quit (Quit: natrys)
23:10:58FromGitter<kaushalmodi> as an aside shashlick and @genotrance sound like different people (I know you are the same :P)
23:11:11FromGitter<kaushalmodi> when I read shashlick, I think of nightlies
23:11:17FromGitter<kaushalmodi> and genotrance reminds of nimterop
23:11:19FromGitter<kaushalmodi> :)
23:12:16shashlickhaha ya its about time i gave up shashlick on irc
23:12:17FromGitter<kaushalmodi> shashlick: it's just the removal of double quotes?
23:12:29shashlickno, it isn't caching at all
23:12:35FromGitter<kaushalmodi> actually it's prudent to use double quotes whenever in doubt
23:12:51shashlicki am trying by removing $TRAVIS_BUILD_DIR from caching let's see
23:13:07shashlickbut also added travis folding so we can actually go thru everything
23:14:12*craigger joined #nim
23:17:24*craigger_ joined #nim
23:18:56*craigger quit (Ping timeout: 250 seconds)
23:20:03*craigger_ quit (Client Quit)
23:20:07*Vladar quit (Remote host closed the connection)
23:22:03*craigger joined #nim
23:39:14*craigger quit (Quit: bye)
23:41:24*craigger joined #nim
23:46:19FromGitter<iffy> Is there a way to run `finish.exe` non-interactively? (i.e. answer yes to everything)
23:46:27FromGitter<iffy> When installing Nim on Windows
23:53:54*craigger quit (Quit: bye)
23:56:20*craigger joined #nim