<< 30-10-2017 >>
00:07:47*Jesin quit (Quit: Leaving)
00:11:34FromGitter<mratsim> Seems like there is a breaking change to `Slice` ⏎ ⏎ ```Error: cannot instantiate Slice ⏎ got: (type int) ⏎ but expected: (T, U)``` [https://gitter.im/nim-lang/Nim?at=59f66e36f7299e8f534a635a]
00:14:17*Jesin joined #nim
00:45:53*Senketsu joined #nim
01:12:29*dddddd quit (Read error: Connection reset by peer)
01:19:47*skrylar joined #nim
01:20:26skrylarinteresting
01:21:09skrylarupdated emacs and such to fiddle around with nim modes again, and the git submodule seems to be causing an excessive amount of redraws whenever editing files in git folders o.O
01:25:57GitDisc<treeform> Do you guys know if tulayang vistis this channel?
01:27:33FromGitter<superfunc> do we have any facility in the macros lib for replace some identifier with another?
01:28:43GitDisc<treeform> maybe they go by the name of yglukhov ...
01:28:44skrylari think you're intended to use unquoting for that
01:35:00*cspar_ joined #nim
01:37:35*cspar__ quit (Ping timeout: 248 seconds)
01:49:23skrylarinteresting graphics question moment: is projection painting inherently slow, or are some of these tools just being weird
01:50:12skrylarFrom what i've gathered, there IS an overhead from finding which triangles are under the cursor and then mapping those back to UV space, but some of that you can get around with readback
01:51:03skrylarguess i'll have to profile what a stack of matrix multiplications actually adds up to
01:55:07FromGitter<gogolxdong> I have put SDL2_net.dll into nim path, compile and run net_server.nim in sdl2 examples shows `could not load: SDL2_net.dll`
02:07:00*Snircle joined #nim
02:18:27*chemist69 quit (Disconnected by services)
02:18:33*chemist69_ joined #nim
02:27:29*UxerUospr joined #nim
02:35:50UxerUosprI'm having problems using the nre library: https://gist.github.com/optik-aper/67419f071a026107879b0d54b07f8f8a
02:36:54UxerUosprI'm looking at the tests code for examples, but even copying and pasting the code from the test yeilds the same compiler type mismatch error
02:37:45*vlad1777d quit (Remote host closed the connection)
02:43:05*gokr quit (Ping timeout: 246 seconds)
02:50:36UxerUosprI opened issue #6629 for my problem
03:02:41*Senketsu quit (Remote host closed the connection)
03:03:51*UxerUospr quit (Quit: Lost terminal)
03:09:11*skrylar quit (Quit: Leaving)
03:10:42*Snircle quit (Quit: Textual IRC Client: www.textualapp.com)
03:17:30FromGitter<gogolxdong> Did I miss anything with SDL2_net.dll?
03:35:15FromGitter<superfunc> skrylar: thanks, I ended up getting by with a simple function and copyChildrenTo
04:06:27GitDisc<treeform> I did play around with SDL2 in nim, but if you are going to do openGL I recommend GLFW its much simpler and better. Unless you are using the slow 2d pixel bliting stuff.
04:54:28*FromGitter * Varriount wonders why no one uses allegro
04:55:58*Senketsu joined #nim
04:56:02*Senketsu quit (Remote host closed the connection)
05:16:44FromGitter<gogolxdong> not much opengl, just need a cross-platform GUI which can send and receive tcp request to interactive with libvirt server, better can run in web GUI form.
05:23:12FromGitter<gogolxdong> @yglukhov Is it feasible? nimx example can be compiled to js and emscripten , I think there is a way like tcp request sent by pushing the button in html5.
05:28:37*vlad1777d joined #nim
05:41:14*vlad1777d quit (Ping timeout: 255 seconds)
05:52:13*vlad1777d joined #nim
06:07:27*Vladar joined #nim
06:41:04*nsf joined #nim
06:52:06watzonIs it possible to turn off garbage collection in the compiler?
07:19:14*gangstacat quit (Quit: Ĝis!)
07:19:17Araqwatzon: koch boot -d:release --gc:none ?
07:23:41watzonThere we go
07:23:44*gangstacat joined #nim
07:24:03*miran joined #nim
07:30:37FromGitter<mratsim> @gogolxdong there was an electron-nim project/proof of concept
07:36:53*PMunch joined #nim
07:47:21*tdc joined #nim
07:49:06*tdc_ joined #nim
07:52:43*tdc quit (Ping timeout: 258 seconds)
07:56:43*Arrrr joined #nim
07:56:43*Arrrr quit (Changing host)
07:56:43*Arrrr joined #nim
07:57:27FromGitter<gogolxdong> @mratsim thanks ,I think it's helpful and it's from PMunch ,this is great.
07:57:53PMunchHuh?
07:58:18FromGitter<gogolxdong> jester+karax+nim-electron
07:59:12PMunchAh :) nim-electron is very minimal by the way
07:59:20PMunchI was more or less just testing if it would work
08:03:39*dddddd joined #nim
08:05:15*xet7 quit (Quit: Leaving)
08:07:06FromGitter<gogolxdong> You provide me timely help.
08:08:16FromGitter<mratsim> I’ve got an unexplainable openmp regression instead of being 8x faster than single-threaded it’s slower (even 5x slower sometimes)… Tried: ⏎ ⏎ 1) Changing Nim version —> nothing ⏎ 2) git bisect —> nothing ⏎ 3) reverting to GCC-6 helps somewhat but only by 20% ... [https://gitter.im/nim-lang/Nim?at=59f6ddf0d6c36fca31b30e15]
08:10:29Araqwell ... if changing the Nim version doesn't help
08:10:42Araqwhat are we supposed to do?
08:11:47FromGitter<mratsim> I wasn’t asking for anything. Trying to check if OpenMP 5.0 introduced a regression somehow ...
08:16:49*Sembei joined #nim
08:18:09*Pisuke quit (Ping timeout: 248 seconds)
08:20:15FromGitter<Yardanico> @mratsim about slices - yes, slices now contain index and backwards index
08:20:35FromGitter<Yardanico> You can read the change it changelog.md
08:20:39FromGitter<Yardanico> *in
08:28:09*yglukhov joined #nim
08:32:21*yglukhov quit (Ping timeout: 240 seconds)
08:34:07Arrrrthe ^1 is mostly used for seqs and strings right?
08:35:08Araqyes
08:36:42FromGitter<mratsim> Seems like the OpenMP “regression/slowdown” only happen on Mac at a specific commit. Can’t reproduce it on Linux...
08:36:46*Ven joined #nim
08:37:09*Ven is now known as Guest36397
08:41:33*gokr joined #nim
08:46:16*yglukhov joined #nim
08:52:17*JappleAck joined #nim
08:58:12*floppydh joined #nim
09:03:27*Sembei quit (Ping timeout: 240 seconds)
09:03:39*MyMind joined #nim
09:09:06*slim_jim_jiss joined #nim
09:12:49*Guest36397 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
09:13:56*claudiuinberlin joined #nim
09:14:40yglukhovAraq: hi, varargs doesn't work with roof op.
09:15:07Araqshould it?
09:15:25yglukhovused to work. i think it should, why not?
09:15:34yglukhovits almost like an openarray
09:15:55Araqdoes it work for openarray?
09:16:10yglukhoverr... let me check
09:17:21yglukhovAraq: it doesn't
09:17:25yglukhov(
09:17:54yglukhovbut it should, right? =)
09:18:04AraqI guess...
09:18:27AraqI think the versions that currently take 'seq' could use 'openArray' instead
09:20:47yglukhovand array versions will be preferred over openarray, right?
09:21:04Araqyup
09:21:40Araqbtw the breaking language changes have come to an end, I think
09:21:52Araqonly stdlib cleanups remain for v1
09:22:14yglukhovAraq: another problem. addr myArr[^1] doesn't work now
09:22:47Araqargh :-)
09:22:58Araqso we need a 'var' overload for that one, oh my
09:23:30yglukhovyup, and a bunch of tests i think =)
09:24:21yglukhovbtw, var would not fix it completely, because unsafeAddr will still not work
09:24:47yglukhovAraq: maybe `[]`(BackwardIndex) should be templates, not procs?
09:25:04Araqcan you work on that?
09:25:24*NimBot joined #nim
09:25:52yglukhovnot in the nearest time im afraid. lots of stuff to handle after my vacation.
09:26:37Araqalright, fixing it then
09:26:52yglukhovbtw, what do you think of https://github.com/nim-lang/Nim/pull/6614
09:26:54Araqbut er
09:26:58yglukhovisnt that genious?
09:27:19Araqthey can't be templates because that would reintroduce the very bug I fixed
09:27:51Araqok, maybe with unsafeAddr hacks but then the JS backend suffers...
09:28:53yglukhovAraq: they could use tmp vars, copy all arguments to those
09:29:07yglukhovor... actually, no..
09:29:36Araqyglukhov: well I don't really understand it, so yeah, it's genious
09:30:04yglukhovAraq: can you ping cheatfate to review it?
09:30:19Araqactually addr myArr[^1] doesn't have to work
09:30:33yglukhovoh really? =))
09:30:40Araqit's clearly a breaking change how ^1 works
09:30:59Araqand the error message is honest about it, so meh
09:31:03yglukhovyou could get away with that, but thats cheating =)
09:31:41yglukhovalso that looks intuitively inconsistent, dont you think?
09:33:14yglukhovafk 15m
09:34:47PMunchAraq, I tried to ask dom96 about this but he told me to ask you. I was wondering if there was a particular reason why there is no way to explicitly call a template or macro while parsing AST.
09:35:40PMunchI managed to get something working by using getAst and expandMacros, but it was evaluated in the scope of the outer macro implementation, which obviously didn't work.
09:39:15Araq"intuitively inconsistent", sure whatever, I'm after consistency after you've learned the basics of the freaking language
09:39:59AraqPMunch: what do you mean?
09:40:43PMunchLet's say I have a macro like for-example the routes macro in jester
09:41:44PMunchWhich takes mostly code but with some extra symbols mixed in. Now someone wants to create a template to for example create a full REST for a resource.
09:42:21PMunchProblem is that templates within the macro aren't evaluated until after the macro has run. So the macro is unable to do what's needed with the symbols the template would spit out
09:43:15PMunchIt would be nice if we could, after having found a template call node in some AST, call a function to expand that node to the output of the template.
09:43:26*slim_jim_jiss quit (Ping timeout: 246 seconds)
09:44:24Araqand how could the macro evaluate the template? templates can be overloaded and type information is not yet available
09:45:44PMunchHmm, good point
09:47:07PMunchBt wouldn't Nim be able to do type-checking on the arguments?
09:47:27PMunchhttps://github.com/dom96/jester/issues/34 This is the issue that got me thinking about this
09:47:28FromGitter<mratsim> I've checked on OpenMP forums, seems like OpenMP thread creation is very costly on OSX: http://forum.openmp.org/forum/viewtopic.php?f=3&t=1583 http://forum.openmp.org/forum/viewtopic.php?f=3&t=1017
09:48:12PMunchI've run into this kind of problem before, and I feel like Nim should have some way of doing this.
09:50:17*zolk3ri joined #nim
09:51:24AraqPMunch: it comes up all the time
09:51:33Araqbut I dunno how to solve it
09:56:14yglukhovAraq: so to summarize: roof should work with openarrays and varargs, but unsafeAddr should not work with roof? addr could probably work with some effort...
09:56:49Araqroof should work with openarrays and varargs
09:57:13yglukhovthats it? =)
09:57:33AraqI don't care about addr that much, I need to see how many different overloads that requires
09:58:11Araqwhy does thttpclient hang on my local machine?
09:58:21Araqit works in travis.
09:58:28Araqfirewall?
09:58:53yglukhovunlikely...
10:01:08yglukhovAraq: why 2 params for Slice?
10:01:30*thomasross__ joined #nim
10:01:42Araqso that 0 .. ^1 can be constructed?
10:02:00Araqbbl
10:02:51yglukhovAraq: i'm not sure what to think of it... there seems to be no way of making the code compatible with 2 nim versions at once. meaning thats pretty hard to trnasition...
10:03:41*thomasross_ quit (Ping timeout: 240 seconds)
10:04:12*slim_jim_jiss joined #nim
10:10:57FromGitter<mratsim> Same issue, I have to choose devel or 0.17.2 now ...
10:22:15FromGitter<gogolxdong> ``````
10:23:02*chaosdav quit (Ping timeout: 260 seconds)
10:23:24FromGitter<gogolxdong> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=59f6fd9cb20c64242965922f]
10:25:38FromGitter<gogolxdong> @PMunch works well in windows but no in linux ubuntu. Have followed the instructions ,not sure where to look into.
10:26:36floppydhsomehow struggling to get a simple unix-domain connection with lib/net.nim - does someone have any demo code for me by any chance?
10:29:24Arrrrasync?
10:30:13floppydhis there an async implementation for them?
10:30:43floppydhas it happens I found my answer for blocking ones: https://github.com/nim-lang/Nim/issues/3320 :)
10:30:48floppydhright after I wrote here
10:32:06*Trustable joined #nim
10:39:49yglukhovAraq: besides addr, you can't assign to memebers. myArr[^1].a = 123
10:40:51yglukhovAraq: how about reverting all of this, and trying to rethink the problem once again. Slices change is particularly harmful. there should be some way to transition gracefully.
10:41:47yglukhov@mratsim, what do you think?
10:45:34PMunchgogolxdong that's strange. I wrote it on a Linux machine and never tried it on Windows :P
10:45:41PMunchSo I would expect it to be the other way around
10:45:55floppydhcan I manually grow a sequence ?
10:46:07PMunchfloppydh, setSize?
10:46:40PMunchErr, setLen
10:47:07floppydhPMunch: thanks a lot
10:49:02FromGitter<gogolxdong> yeah,it's strange ,I though the same as you.
10:49:04PMunch@gogolxdong could you show me the contentt npm-debug.lo file?
10:49:45PMunchUgh, I've managed to somehow enable insert mode and I don't have a key to turn it off..
10:50:47FromGitter<gogolxdong> http://paste.openstack.org/show/624951/
10:52:06floppydhhow do I get the address of the first element of a sequence Foo again?
10:52:33PMunch@gogolxdong, can you make sure that the original quickstart thing works?
10:52:33Arrrraddr(foo[0])
10:53:08FromGitter<gogolxdong> it's strange ,the version in windows in ~1.6.2 ,in linux it's 1.6.5.
10:54:27FromGitter<gogolxdong> but I changed both to ~1.7.8, windows works well too
10:56:28FromGitter<mratsim> @Araq is there any difference in Nim Allocator between Linux and macOS, some OpenMP pattern may trigger either memory fragmention
10:57:00FromGitter<mratsim> https://stackoverflow.com/questions/5875989/does-multithreading-emphasize-memory-fragmentation
10:57:08FromGitter<mratsim> Or "false sharing"
10:57:32FromGitter<mratsim> https://stackoverflow.com/questions/6605677/openmp-poor-performance-of-heap-arrays-stack-arrays-work-fine
10:57:47FromGitter<gogolxdong> electron-quickstart package can't work straight well.
11:02:16*Snircle joined #nim
11:05:44FromGitter<gogolxdong> in both platform.
11:07:43floppydhI can't take the addr of an argument no?
11:08:33FromGitter<mratsim> You can if it's car or a ref type
11:08:38floppydhcurrently copymeming values in a generic proc into a seq[char], but then I can't pass like string literals etc to the proc, has to be a var with an address, I assume one can do the same with value semantics? - like adding the char-representation of any type whatsoever
11:11:09FromGitter<mratsim> Use unsafeAddr s[0]
11:16:38floppydhmratsim thanks a lot
11:18:59FromGitter<gogolxdong> ubuntu distribution 17.10
11:23:04*tdc_ quit (Read error: Connection reset by peer)
11:23:29*tdc_ joined #nim
11:25:05*yglukhov quit (Remote host closed the connection)
11:25:39*yglukhov joined #nim
11:27:07*yglukhov quit (Remote host closed the connection)
11:27:19*yglukhov joined #nim
11:31:13Araqmratsim: there is no difference in implementation but different mmap() behaviour can also cause it
11:31:25yglukhovAraq: ok, var overloads seem to work...
11:31:34Araqyglukhov: I can make it compatible with the old version I think
11:31:46yglukhovSlices?
11:31:48yglukhovhow?
11:32:10yglukhovsorry, battery is low. will go offline now.
11:32:11AraqSlice[T] should stay but an alias for HeterogenousSlice[T, T]
11:32:22Araqor something with a better name...
11:32:30yglukhovyup. might be good.
11:32:59yglukhovbbl
11:33:15*yglukhov quit (Remote host closed the connection)
11:33:36FromGitter<mratsim> @Araq, thanks, I guess I'll just say, if you can't eat your Apple and have high performance too *shrug*
11:33:40*yglukhov joined #nim
11:38:17*yglukhov quit (Ping timeout: 260 seconds)
11:38:52*elrood joined #nim
12:02:49*yglukhov joined #nim
12:12:42FromGitter<dandevelo> I am trying to have some handlers (procs) and being able to get them by name. So I thought of creating a Table[string,proc]. Are there any other options?
12:17:11floppydhtrying to pass a var to a template/macro and inside print both the name of the variable and its repr/contents - any pointers?
12:20:17*jjido joined #nim
12:31:13ArrrrastToStr
12:31:36Arrrrdandevelo that's probably the best option
12:32:20ArrrrA faster approach would be to use arrays and enums, but depends on your requirements
12:33:29FromGitter<dandevelo> How would the arrays and enums approach work?
12:34:11Arrrrarray[MyEnum, proc]
12:34:46FromGitter<dandevelo> That would work with a predefined list of functions, right? Not with variable function names
12:34:49FromGitter<Yardanico> yeah
12:34:50ArrrrFor every proc you add you make sure to add a new enum value
12:36:14FromGitter<dandevelo> Yep. Thanks!
12:41:40FromGitter<mratsim> Note: identifier construction is very useful to create proc with variable names
12:42:05FromGitter<dandevelo> @mratsim not sure I understand
12:42:08FromGitter<mratsim> https://nim-lang.org/docs/manual.html#templates-identifier-construction
12:42:46FromGitter<dandevelo> Nice! Didn't know about those
12:42:49FromGitter<dandevelo> Thanks!
12:42:54FromGitter<mratsim> ```template createProc(name: untyped) = proc `name`() = discard```
12:43:32*Jesin quit (Quit: Leaving)
12:46:02FromGitter<dandevelo> When using a table, is there a way to check if a key is inside the table and get the value at the same time? I know that [] will raise an exception if the key doesn't exist. Is that also the case for release mode?
12:47:49*Trustable quit (Remote host closed the connection)
12:56:59Araqdandevelop: use getOrDefault
12:57:02*gokr quit (Ping timeout: 260 seconds)
12:59:31FromGitter<Yardanico> @dandevelo yes it will raise an exception in release mode too
12:59:36FromGitter<Yardanico> exceptions aren't disabled in release mode
13:00:18FromGitter<dandevelo> Thanks @Yardanico not sure why I got that impression but it was from a discussion I had here a few weeks ago
13:01:49floppydhso am I going to get hanged by the neck if rather then wrapping a whole big c-lib to use a small bit of it, I rather just write a small c file that exposes that functionality and then import a few of those interface functions as nim procs?
13:02:33Araqwrite a small Nim file instead
13:02:49floppydhAraq: I'd have to emit tho?
13:04:01floppydhI'll try that first, the multi line string-syntax for emit seems a bit unwieldy, but sure
13:04:23floppydhI see `foo` is just supported for standard string literals now
13:04:53floppydhoh, or is that the point, to encourage people to import them ?
13:07:41*smt joined #nim
13:10:33*smt` quit (Ping timeout: 246 seconds)
13:14:59Araqfloppydh: what do you mean? `foo` is supported like it always was
13:15:11floppydhAraq: manual says only for "..." not for """..."""
13:15:16floppydhand that its deprecated I think
13:15:20floppydhfor emit
13:15:49Araqwell yes, emit works better with the list of items
13:15:49floppydh" For backwards compatibility, if the argument to the emit statement is a single string literal, Nim symbols can be referred to via backticks. This usage is however deprecated. "
13:15:58floppydhah okay
13:16:06Araqhmm that seems wrong
13:16:19Araqthe backticks are for backwards compat but they work in any string literal
13:16:32floppydhoh
13:16:54floppydhbut I assume I can either write a nim-proc then emit C in it, or just top-level-emit a c function and then import it as a nim-proc?
13:16:59floppydhwhat is preferred?
13:18:17Araqwrite a Nim proc instead without 'emit'
13:18:41Araqbut if you really need the emit, make it as small as possible
13:19:41floppydhgotcha
13:20:13*UxerUospr joined #nim
13:21:36UxerUosprI found a typo in the manual on: https://nim-lang.org/docs/manual.html#converters
13:22:03FromGitter<SitiSchu> Do a PR ^^
13:22:22UxerUosprOkay
13:24:45PMunchWait, what's the typo?
13:25:04UxerUosprthe anchor for convertible relation
13:25:14UxerUosprdoesn't exist
13:25:29FromGitter<Yardanico> yeah, change it to https://nim-lang.org/docs/manual.html#type-relations-convertible-relation
13:25:29Araqalready fixed in devel
13:25:34AraqI think
13:25:35UxerUosprand should be https: #type-relations-convertible-relation
13:25:55PMunchAaah, right. Yeah a lot of anchors are broken..
13:25:59UxerUosprs/https//g
13:26:30UxerUosprAraq: thanks!
13:26:33*yglukhov quit (Remote host closed the connection)
13:30:50FromGitter<Yardanico> UxerUospr: yeah, it's fixed
13:31:21UxerUospr@Yardanico: cool, thanks!
13:32:28*yglukhov joined #nim
13:41:01FromGitter<gogolxdong> @PMunch sovled. rm node_modules/electron then npm install electron -g for linux
13:42:02PMunchAah, good to hear you figured it out
13:42:13AraqPMunch: what do you think about this:
13:42:30Araqintroduce 'semityped' as a macro parameter mode
13:42:46Araqthat type checks the body but doesn't produce errors
13:42:59Araqthe macro can then "fix" the wrong parts
13:43:05FromGitter<Yardanico> hmmm
13:43:08Araqand sem'check runs completely afterwards
13:43:25Araqso a macro naturally sees template expansions
13:43:35PMunchOoh, yes that sounds great!
13:43:45Araqbut needs to be able to deal with partially sem'checked ASTs
13:44:01Araqthough that's usually not much work to do in practice
13:44:43PMunchMost of the macros that do these replacement things typically uses untyped anyways since the things it want to replace doesn't really exist
13:45:57Araqyes.
13:46:02FromGitter<Yardanico> but I really hope it would be properly documented
13:46:08Araqit'll be really tricky to implement though
13:46:16FromGitter<Yardanico> so all can understand differences between untyped/semityped/typed
13:47:00Araqyardanico: maybe we should just change what 'typed' does?
13:47:23Araqit's still 'typed', it's just that errors are detected later and don't prevent macro invokations
13:47:39PMunchYeah, that actually surprised me the first time I used typed
13:47:54Araqbut implementing this will be a pita, the compiler is written with "continue after error" in mind
13:48:02Araqand does AST error corrections
13:48:06Araqbut still ...
13:48:41Araqanother HUGE "attack surface"
13:48:57PMunchI think it's something that will really improve the usability of macros and DSTs
13:49:12Araqagreed, right now it's hardly composable
13:49:34Araqwill also improve nimsuggest support within macro bodies as a side effect
13:50:06PMunchWhat do you mean by "attack surface" in this context?
13:50:46Araqit's a combinatorial explosion to test against, every node kind comes in 2 flavours: errornous and correct
13:51:15Araqand things like nkVarSection actually change the shape after sem'check
13:51:32Araqthough that's probably always been a bad idea to begin with and needs to wiped out
13:54:24*Jesin joined #nim
14:03:34*Trustable joined #nim
14:05:29*Arrrr quit (Read error: Connection reset by peer)
14:06:45PMunchHmm, yeah I can see that being a problem..
14:07:01PMunchBut I think it's not going to cause too many issues
14:12:37*Jesin quit (Quit: Leaving)
14:16:34*yglukhov quit (Remote host closed the connection)
14:16:47Araqlike my "changed how ^ is implemented" ?
14:17:06AraqI too thought "it's not going to cause many issues"
14:17:24*JappleAck quit (Quit: Leaving)
14:18:32PMunchHaha, fair point
14:19:12PMunchWell, if it's implemented as a new semityped type then it at least won't break old code
14:19:29PMunchWhether or not it is justified to have both is another issue
14:20:38PMunchI'd say maybe implement it as a new type first, then if it turns out to be too little difference for it to be useful to differentiate then they can me merged.
14:23:16AraqI don't see why, macros that right now take 'typed' continue to work
14:23:28Araqsince right now only valid code can be passed to them
14:23:34Araqwell ok
14:23:44Araqthey secretly need to support a new AST scheme, hmmm
14:26:13PMunchHmm, that is a valid point. However if I were to change something while working on the devel version it could break things for people using a "stable" version
14:29:58FromGitter<Yardanico> That's why you can have branches on github:)
14:30:08FromGitter<Yardanico> Devel and master
14:30:38FromGitter<SitiSchu> branches on git* :P
14:31:02*Jesin joined #nim
14:31:18FromGitter<Yardanico> Yeah, story
14:31:22FromGitter<Yardanico> Sorry :D
14:33:30*wishi quit (Remote host closed the connection)
14:35:44*wishi joined #nim
14:39:13FromGitter<data-man> @Araq: HSlice? :) ⏎ it would be nice if it was possible to write so: ⏎ ⏎ ```type ⏎ Slice*[T, U] = object ⏎ a*: T ⏎ b*: U ⏎ Slice*[T] = Slice[T, T]``` [https://gitter.im/nim-lang/Nim?at=59f73991210ac269204c14dd]
14:42:02Araqsure, overloading generics based on arity
14:42:53AraqI'll happily support every useful feature for version 1 cause that's how to get software quality. :P
14:43:25*gokr joined #nim
14:48:28FromGitter<Yardanico> Araq: maybe you can use github milestone thing so we all can see what will be fixed/implemented before 1.0 ? :)
14:48:38FromGitter<Yardanico> https://github.com/nim-lang/Nim/milestones
14:55:12Araqyardanico: the last time we tried to use a "milestone" didn't go well
14:55:28Araqso nah, better keep my todo.txt up to date
15:03:14*miran_ joined #nim
15:19:02*Ven joined #nim
15:19:46*Ven is now known as Guest42207
15:27:56*PMunch quit (Quit: Leaving)
15:49:06*nsf quit (Quit: WeeChat 1.9.1)
15:50:50FromGitter<data-man> @Araq: Are you here now? :)
15:53:10*JappleAck joined #nim
15:53:47Araqsure
15:54:58FromGitter<data-man> I already asked before, but what do you think about https://github.com/blog/2392-introducing-code-owners? ⏎ In the D language, relatively recently, this was done for Phobos. ⏎ ⏎ Do you plan to expand the Nim development team? ⏎ IMHO, this would the Nim benefit. [https://gitter.im/nim-lang/Nim?at=59f74b51976e63937e064b16]
16:04:51Araqwe need to expand the development team, fully agree with that
16:05:16*byte512 quit (Read error: Connection reset by peer)
16:06:23Araqbut we can't just make up more rules for contributors, contributing needs to stay as simple as possible for now in order to attract new core developers
16:08:03Araqit's fine if you say "I am adding xyz to the stdlib and will continue to maintain it" but it requires trust, so ... we need to be sure you'll here for some time
16:08:48*byte512 joined #nim
16:10:00Araqdo you want to become a core dev with write access? :-)
16:11:46*gokr quit (Ping timeout: 264 seconds)
16:12:15Araqbtw it is my impression D's development is slowed down by a too restrictive process and it's hurting D a bit. I'm following its forum quite a bit and not many are willing to go through the ultra-hard review process.
16:12:55Araqcould be wrong, ofc
16:16:22FromGitter<data-man> Oh yeah! I would really want to become a core dev with write access! :) ⏎ But I'm with the Nim too short to claim it. ⏎ I'm very humble. :)
16:17:21Araqwell I see you're active with the PRs, that's nice
16:17:47Araqcontinue on this path and I'll make you a core dev within months ;-)
16:18:07Araqtrust takes time.
16:21:13*smt` joined #nim
16:21:59FromGitter<data-man> Thank you! ⏎ But I'm probably too old for that. ⏎ Maybe I'm the oldest here. :) ⏎ I was born when computers were large and chewed punch cards, and a memory of 1 megabytes was a luxury. :) [https://gitter.im/nim-lang/Nim?at=59f751a732e080696e4372ce]
16:23:47Araqer ... should I be concerned now? you don't intend to live for a couple of months?
16:24:22*smt quit (Ping timeout: 255 seconds)
16:27:42FromGitter<data-man> Oh, no! ⏎ I hope to stretch out a couple more ... ⏎ ... tens of years. :)
16:30:22*sleepyqt joined #nim
16:39:31*yglukhov joined #nim
16:48:59*Guest42207 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
16:51:00*nsf joined #nim
16:51:02FromGitter<UrKr> `import module as module_alias` precludes calling function from module with this notation `value.f(x)` right?
16:52:05FromGitter<Varriount> @UrKr I believe that's when you do "from ... Import nil"
16:54:07*Ven joined #nim
16:54:31*Ven is now known as Guest45092
16:55:20*Jesin quit (Quit: Leaving)
16:55:57FromGitter<UrKr> Right, you mean import the function directly? ⏎ Any opinions on `value.module_alias.f(x)`
16:56:12*Guest45092 is now known as Ven``
16:57:02FromGitter<Varriount> I don't believe that's possible.
16:59:31FromGitter<UrKr> It's not - I tried :) just wondering if it's something to be considered, since the now the `x.f(y)`notation has to be refactored if you run into a nameclash for `f`, right?
17:00:21*PMunch joined #nim
17:01:02Araqyeah but it has to be "refactored" / edited anyway
17:01:20Araqmoving the 'x' into the 'f(' is not that much work
17:03:52FromGitter<data-man> @Araq: Do you plan for version 1.0 to add a loop with a postcondition? (do / while as in the C (JS) or repeat / until as in the Pascal) ⏎ This is one of fundamental programming constructs, and I was very surprised that this is not in Nim. ⏎ In addition, for such a loop there is a direct analogue in the C and JS.
17:04:41Araqnah, it's a template/macro away and Python lacks it too.
17:05:13Araqit's also not obvious what it should look like given Nim's indentation based syntax
17:05:17Araqwhile true:
17:05:21FromGitter<UrKr> Hm... yeah you're right I suppose. Still something like `s.f(x).g(y).h(z)` would be slightly more annoying. Also, you're right "edited" is more appropriate here... "refactor" sounds a bit affected
17:05:24Araq if cond: break
17:06:03FromGitter<UrKr> Hm... yeah you're right I suppose. Still something like `s.f(x).g(y).h(z)` would be slightly more annoying. Also, you're right "edited" is more appropriate here... "refactor" sounds a bit affected
17:06:13AraqUrKr: that dot notation causes some trouble already, it's worth it to unify procedural and OO concepts
17:06:21Araqbut shouldn't be pushed any further IMO
17:06:38*dddddd quit (Remote host closed the connection)
17:07:27FromGitter<UrKr> Right, I suppose I agree.
17:08:02Araqin fact, I thought about distinguishing between x.field() and (x.field)()
17:08:09Araqin order to make it less ambiguous
17:08:41Araqbut that doesn't solve much so I intend to keep things as they are
17:11:50AraqUrKr: btw to resolve ambiguities usually 'import x except clash, clashB' is usually the better idea
17:15:52FromGitter<data-man> ```i = 0 ⏎ until i < 10: ⏎ i.inc``` ⏎ ⏎ ??? [https://gitter.im/nim-lang/Nim?at=59f75e48d6c36fca31b5f187]
17:18:09*Jesin joined #nim
17:21:56FromGitter<vberset> Hello. I want to write to file chunks of data while I compute some values based on them, asynchronously. I try to achieve that using the `AsyncFile` API. But, it doesn't expose a function to flush the OS buffer. This would have allowed me to be sure at the end of the chunk's processing that the (untouched) chunk is effectively written on disk. Is there a proper solution to achive that with `AsyncFile`? Or should I use
17:21:56FromGitter... other primitives?
17:25:02*floppydh quit (Quit: WeeChat 1.9.1)
17:25:28*yglukhov quit (Remote host closed the connection)
17:26:01*jjido quit (Ping timeout: 255 seconds)
17:44:18dom96vberset: easiest thing to do might be to add a `flush` into asyncfile
17:53:16*yglukhov joined #nim
17:55:18*PMunch quit (Quit: leaving)
17:57:27*yglukhov quit (Ping timeout: 240 seconds)
18:01:12*ofelas joined #nim
18:01:29FromGitter<stisa> @data-man the template would look something like https://play.nim-lang.org/?gist=15d3b99149747490a5b6de3fd74042aa
18:11:04*tdc_ quit (Read error: Connection reset by peer)
18:11:21*jjido joined #nim
18:11:50*tdc joined #nim
18:12:59*captainkraft quit (Quit: WeeChat 1.9)
18:16:23*Vladar quit (Remote host closed the connection)
18:17:46*jjido quit (Ping timeout: 255 seconds)
18:17:56*jjido joined #nim
18:22:00FromGitter<data-man> @stisa: Thank you! ⏎ I was hoping that a macro can do almost anything, but I'm thinking about the generated machine code. ⏎ First I'll do some speed research. :)
18:23:20*nsf quit (Quit: WeeChat 1.9.1)
18:24:00Araqthe machine code has no inefficiencies, it's basic control flow stuff
18:24:19Araqthat was close to perfect in the 80ies
18:25:12Araqand so nobody suggested "replace while loop with goto for speed" in C back then
18:28:34*durandu joined #nim
18:30:26*durandu quit (Client Quit)
18:31:05*yglukhov joined #nim
18:31:06*ofelas quit (Quit: shutdown -h now)
18:35:43*yglukhov quit (Ping timeout: 248 seconds)
18:40:43*yglukhov joined #nim
18:44:30*yglukhov quit (Read error: Connection reset by peer)
18:45:06*yglukhov joined #nim
18:46:07*rauss quit (Quit: WeeChat 1.9.1)
18:47:28*rauss joined #nim
18:58:06*vivus joined #nim
19:00:15*miran_ quit (Quit: Leaving)
19:06:09*Ven`` quit (Read error: Connection reset by peer)
19:14:50*slim_jim_jiss quit (Ping timeout: 255 seconds)
19:28:36*lurker joined #nim
19:29:37*Sentreen quit (Ping timeout: 260 seconds)
19:35:17*xet7 joined #nim
19:39:17GertmCan Nim be compiled on a RPi? (I have the first gen)
19:41:25FromGitter<Yardanico> yeah
19:41:29FromGitter<Yardanico> if you have swap :)
19:42:12*Sentreen joined #nim
19:49:46FromGitter<Varriount> It'll take quite a while though
19:49:51FromGitter<gokr> Nim worked great on rpi3
19:51:02FromGitter<Gooseus> whats the easiest way to read piped data into a string in a nim program? I feel I'm missing something obvious here
19:52:48FromGitter<Varriount> read from the stdin file handle?
19:53:09FromGitter<Varriount> https://nim-lang.org/docs/system.html#stdin
19:53:51FromGitter<Varriount> @Gooseus What are you piping?
19:57:53*Jesin quit (Quit: Leaving)
20:00:58*slim_jim_jiss joined #nim
20:02:12FromGitter<Gooseus> yeah, I'm just reading in data from another command line program... when I tried stdin.readAll I got the data in the var I think, but then I can't echo to stdout and even when I did stdout.write it just writes out the stdin, even though I'm writing other data to stdout: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=59f78544b20c64242968b921]
20:02:27FromGitter<Gooseus> well that didn't format how I wanted at all...
20:03:21*nsf joined #nim
20:07:42yglukhovGooseus: should not you use the pipe operator | instead of file redirection >
20:07:56yglukhovor is that some shell i'm unaware of =)
20:08:25FromGitter<Gooseus> jeez, I did not sleep much last night... ok, derp
20:11:04*xet7 quit (Quit: Leaving)
20:11:15FromGitter<Gooseus> thanks yglukhov, that little bit of stupid cost me way too much
20:11:57yglukhovyeah, programmers need to sleep sometimes. ;)
20:12:46*xet7 joined #nim
20:22:49*Ven joined #nim
20:23:12*Ven is now known as Guest9063
20:23:39GitDisc<treeform> Hey yglukhov, you did the async mysql thing?
20:24:00yglukhovtreeform: nope
20:24:25GitDisc<treeform> ok sorry to bother you
20:24:26yglukhovcassandra, rethinkdb, mongo - maybe. mysql - nope =)
20:24:27*vlad1777d quit (Ping timeout: 260 seconds)
20:24:40yglukhovno problem
20:29:45*tdc quit (Read error: Connection reset by peer)
20:30:20FromGitter<Gooseus> @dom96 @Varriount alright, here is my latest attempt - a nimaws module with sigv4 and awsclient and a service specific s3client - https://github.com/Gooseus/nimaws
20:30:48*PMunch joined #nim
20:30:51FromGitter<Gooseus> I didn't try to publish it to nimble yet, but I setup the .nimble file and structure how I think it's supposed to be, figured I'd get some feedback first
20:31:36FromGitter<Gooseus> gotta go for a walk, will cycle back in a little bit
20:31:39*tdc joined #nim
20:31:43yglukhovGooseus: have you seen this https://github.com/aidansteele/aws_sdk.nim
20:33:16FromGitter<Varriount> @Gooseus Thanks! I'll have a look at it later
20:33:52*gokr joined #nim
20:34:02FromGitter<Varriount> There are some stylistic suggestions I might make, as well as some string optimizations.
20:38:47FromGitter<Varriount> @Gooseus Thanks for all the effort you've put into this... perhaps the possibility of an AWS library for Nim isn't so far-fetched.
20:39:04FromGitter<Varriount> (It also means I might have the possibility of using Nim at work!)
20:45:48*gokr quit (Ping timeout: 240 seconds)
20:45:57*tdc quit (Ping timeout: 240 seconds)
20:47:19*gokr joined #nim
20:52:16*UxerUospr left #nim (#nim)
21:15:30FromGitter<Gooseus> yglukhov, yeah, I was using that as a reference and took a few hints... I mostly started this as an academic pursuit and a potential tool for some basic work stuff if I could get it working
21:22:12FromGitter<Gooseus> but now that it's built and I'm still having fun, I wouldn't mind continuing some more... @Varriount thanks! def let me know
21:24:10*PMunch quit (Quit: leaving)
21:27:43FromGitter<Gooseus> does anyone here know the maintainer of that aws_sdk? there hasn't been any activity in 3 months and it looks like they were also a nim noob, would be good to know if were still planning to work on it
21:30:26*lurker quit (Quit: Leaving)
21:32:05FromGitter<Varriount> @Gooseus I believe he's in the gitter and/or IRC room
21:33:16GitDisc<treeform> Gooseus, "insanely detailed and tedious process." That is no joke. Thank you for doing this.
21:37:56FromGitter<Gooseus> treeform np, had a lot of fun with it despite the AWS nonsense... though I'm going to be revising that shit again and again if I keep working on it and I imagine it'll become much less fun
21:38:19*couven92 joined #nim
21:38:50FromGitter<Gooseus> knowing my luck AWS will deprecate v4 next year
21:39:02*yglukhov quit (Remote host closed the connection)
21:39:13*Jesin joined #nim
21:39:37*yglukhov joined #nim
21:39:55*gooseus joined #nim
21:43:52*yglukhov quit (Ping timeout: 252 seconds)
21:44:51FromGitter<Gooseus> @Varriount I don't see him if he's going by @aidansteele
21:45:08FromGitter<Varriount> I believe he goes by something different.
21:45:13*vlad1777d joined #nim
21:46:28FromGitter<Varriount> @Gooseus Any idea why the process is so convoluted?
21:46:45*yglukhov joined #nim
21:47:30*slim_jim_jiss quit (Ping timeout: 246 seconds)
21:47:43FromGitter<Gooseus> which process?
21:49:24*vivus quit (Quit: Leaving)
21:50:04yglukhovGooseus: the author of aws_sdk doesn't look active, but he actively merged a couple of my prs. i think you could ask him to make you a collaborator if you're concerned.
21:50:31*slim_jim_jiss joined #nim
21:50:34*jjido quit (Remote host closed the connection)
21:50:49yglukhovnot that i'm interested, just want to keep nim community efforts focused =)
21:51:46*claudiuinberlin quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
21:52:21*claudiuinberlin joined #nim
21:52:29FromGitter<Gooseus> @Varriount the sigv4 stuff? Not 100%, but I think it has to do with a combination of how they need to secure and authenticate each request within their iam structure and how a lot of http clients don't guarantee the order of the headers and querystring params
21:53:28FromGitter<Gooseus> yglukhov yeah, I hear ya, I'll reach out
21:59:29*libman joined #nim
22:01:57*gooseus quit (Quit: leaving)
22:05:23*Guest9063 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
22:12:58*slim_jim_jiss quit (Ping timeout: 264 seconds)
22:16:24*JappleAck quit (Quit: Leaving)
22:19:41*claudiuinberlin quit (Quit: Textual IRC Client: www.textualapp.com)
22:24:37*Ven joined #nim
22:25:01*Ven is now known as Guest73995
22:34:35*vlad1777d quit (Ping timeout: 240 seconds)
22:38:44*nsf quit (Quit: WeeChat 1.9.1)
22:53:46GitDisc<treeform> Gooseus, can I feature request to get S3 list working?
22:53:57GitDisc<treeform> for listing files?
22:54:24GitDisc<treeform> I only need list and download and I will be set for life.
23:00:52*sleepyqt quit (Quit: Leaving)
23:05:23*Guest73995 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
23:06:54FromGitter<Gooseus> treeform sure, let's see how hard that is to get going with what I have
23:19:03*yglukhov quit (Remote host closed the connection)
23:22:22*elrood quit (Quit: Leaving)
23:43:34*zolk3ri quit (Remote host closed the connection)
23:55:53*yglukhov joined #nim
23:59:57*yglukhov quit (Ping timeout: 240 seconds)