<< 19-11-2017 >>
00:06:34*BitPuffin|osx joined #nim
00:09:21*couven92 quit (Quit: Client Disconnecting)
00:10:32*nsf quit (Quit: WeeChat 1.9.1)
00:10:40*bozaloshtsh joined #nim
00:10:40*bozaloshtsh quit (Changing host)
00:10:41*bozaloshtsh joined #nim
00:18:28*Snircle_ quit (Read error: Connection reset by peer)
00:24:31*dexterk quit (Ping timeout: 248 seconds)
00:35:24*rbrt joined #nim
00:40:09FromGitter<nitely> literally in hand
00:49:14*rbrt quit (Quit: Oíche mhaith)
00:54:22*rbrt joined #nim
01:15:42laasis there a good resource on macros
01:15:44laas?
01:15:56laasI find the manual and the module-level documentation of macros insufficient
01:17:08*laas quit (Quit: WeeChat 1.9.1)
01:20:07skrylari suppose one could be made
01:25:19*rbrt quit (Quit: Oíche mhaith)
01:27:19*rbrt joined #nim
01:28:08*arnetheduck quit (Remote host closed the connection)
01:29:04*rbrt quit (Client Quit)
01:45:40*tefter joined #nim
01:49:33*jsgrant__ quit (Remote host closed the connection)
02:04:17*rbrt joined #nim
02:19:46*arnetheduck joined #nim
02:56:11*chemist69 quit (Ping timeout: 240 seconds)
03:10:17*BitPuffin|osx quit (Ping timeout: 268 seconds)
03:10:27*chemist69 joined #nim
03:11:05*PMunch quit (Quit: leaving)
03:22:11*redlegion quit (Ping timeout: 255 seconds)
03:27:28*redlegion joined #nim
03:27:29*redlegion quit (Changing host)
03:27:29*redlegion joined #nim
03:36:22*fvs left #nim ("ERC (IRC client for Emacs 25.3.1)")
03:43:12*MJCaley joined #nim
03:46:56*vlad1777d quit (Ping timeout: 255 seconds)
03:58:33*sz0 joined #nim
04:03:09*MJCaley quit (Quit: MJCaley)
04:04:56*MJCaley joined #nim
04:20:45*SenasOzys__ joined #nim
04:23:40*SenasOzys_ quit (Ping timeout: 268 seconds)
04:35:39*MJCaley quit (Quit: MJCaley)
04:43:26*rbrt quit (Quit: Oíche mhaith)
04:53:34*skrylar quit (Remote host closed the connection)
04:58:54*MonsterAbyss quit (Ping timeout: 246 seconds)
04:59:29*MonsterAbyss joined #nim
05:06:41*SenasOzys__ quit (Ping timeout: 248 seconds)
05:07:06*endragor joined #nim
05:22:25*astronavt quit (Quit: Leaving...)
05:55:38*bkerin joined #nim
05:55:57bkerinhiya folks. ima on a quest to turn a string into an identifier in a template. possible?
06:00:49FromGitter<techate> that came up earlier today actually: *<laas>* what does !"some string" mean? *<laas>* what does !"some string" mean? *<Araq>* with extended raw string literals it's not required anymore
06:01:19FromGitter<techate> that repeat was supposed to be: *<Araq>* converts a string into an identifier
06:01:29bkerintechate: I was there for that but sadly I didn't fully understand it
06:01:40bkerinhmm
06:02:50FromGitter<techate> well a grep -Pr ' !"' within nim's repo will find some macros that use it
06:03:12FromGitter<techate> like in nimsuggest/tests/twithin_macro.nim
06:03:42FromGitter<techate> also: https://nim-by-example.github.io/oop_macro/
06:04:54bkerinthe emitted AST for the template when I try that is Prefix->(Sym "!", StrLit foo)
06:05:35bkerinwhat did aray mean about extended raw string lits obsoleting it I wonder
06:08:05*yglukhov joined #nim
06:08:47*brandonwolfhill joined #nim
06:10:27bkerinuntyped/typed/typedesc also confuse the heck outa me
06:11:14*brandonwolfhill quit (Quit: Page closed)
06:12:32*yglukhov quit (Ping timeout: 260 seconds)
06:17:13*dddddd quit (Remote host closed the connection)
06:32:24FromGitter<techate> well I don't think the AST would be special. it's not syntax. it's defined in macros: proc `!`*(s: string): NimIdent {.magic: "StrToIdent", noSideEffect.}
06:36:26FromGitter<techate> of extended raw string literals, I dunno. what I've found on that just explains why echo"\" works and echo "\" is a syntax error
06:39:29bkerinhmm this is one wild wooly lang
06:41:24bkerini think the whole approach of keeping any type info in strings at compile time is wrong or unused and it's all supposed to be in typedesc or something
06:52:56FromGitter<techate> you can avoid !"string" if you want, and use newIdentNode("string").ident , but that doesn't have anything to do with raw string literals
06:56:06*sz0 quit (Quit: Connection closed for inactivity)
06:59:27bkerinboth approaches give NimNode values, which are great for macros but so far as I can tell not quite what I want in this case.
06:59:45*sz0 joined #nim
07:00:50*miran joined #nim
07:01:05FromGitter<techate> maybe that's one of the lines beyond which you need a macro rather than a template
07:01:18FromGitter<techate> what are you trying to do?
07:04:17bkerinsomething like this:
07:04:18bkerintemplate echoVar(varName:string): untyped = echo varName & ": " & `varName`
07:05:22bkerinonly that doesn't work, `varName` just comes out a string, presumably because it's not a maical 'untyped' as in the language manual on Identifier construction in templates
07:06:49bkerinin general templates seem much harder to use than macros the latter are more verbose but its clear how they work templates not so much
07:08:38bkerinthere's also the mysterious {.inject.} pragma in the Identifier construction example which so far as I can find isn't mentioned anywhere else in the documentation
07:11:04bkerinwll bedtime for me thanks for advice, I'll try again tomorrow
07:11:07*bkerin quit (Quit: Visit http://lice.muppetz.com)
07:11:13FromGitter<techate> ah wait a sec
07:11:40FromGitter<techate> https://github.com/nim-lang/Nim/blob/devel/tests/macros/tquotewords.nim
07:12:22FromGitter<techate> a macro that turns quoteWords(some, words, here, without, quotes) into ["some","words","here","without","quotes"]
07:12:28FromGitter<techate> like qw() in Perl or %w() in Ruby
07:13:16FromGitter<techate> I think you'll need a macro rather than a template for that, and that macro does something similar
07:13:21FromGitter<techate> otherwise, g'night
07:13:54FromGitter<techate> for templates being harder, it's easy to do easy stuff :)
07:16:15*endragor quit (Remote host closed the connection)
07:19:06*gokr joined #nim
07:36:08*skrylar joined #nim
07:45:15skrylaris it worth my time to tinker with openmp
07:51:41*endragor joined #nim
07:56:01*endragor quit (Ping timeout: 240 seconds)
07:59:22skrylarit sounds interesting, but i don't know how well it behaves with nim and i think ML stuff runs on opencl these days anyway. dunno
08:03:11Araqtechate: when I introduced `!`, the syntax ident"xyz" was not available
08:03:41Araqand I wanted to avoid excessive () in the macro code
08:04:15Araqthat's all there is to it, `!` is just a prefix operator defined in macros.nim
08:04:25Araqas you noticed.
08:04:38FromGitter<alehander42> that shouldn't matter for newIdentNode, right?
08:05:55Araqident is an alias for newIdentNode
08:06:13Araqoh I see, `!` doesn't even do what I thought it does, lol
08:06:26FromGitter<techate> oh, I didnt' realize ident"str" worked. but yeah: ident"str".ident works in that animals macro example
08:06:52Araqwell I was wrong
08:07:11Araq`!` doesn't even do what I thought it does
08:08:26FromGitter<alehander42> hygiene?
08:28:01Araq`!` is so rare that I forgot what it means
08:28:36Araqwe use operators for common operations though, not for fringe cases so `!` needs to be replaced with toNimIdentifier
08:29:25*yglukhov joined #nim
08:46:16*claudiuinberlin joined #nim
08:47:47FromGitter<data-man> Good day! ⏎ @Araq: Why are a concepts almost not used in the stdlib?
08:47:50FromGitter<mratsim> @skrylar OpenMP is realtively easy for map/reduce/fold. And sorel lacking from Julia and Numpy (don’t know about matlab). There are a lots of preprocessing in ML that could benefit from parallelization but lacks Cuda implementation. (all the stuff on Scikit learn)
08:48:29*Vladar joined #nim
08:52:42skrylarmratsim: ah. i have little interest in cuda.
08:52:54skrylari would be willing to learn OpenCL, but it seems weird to get in to
08:54:56FromGitter<mratsim> There is a tutorial to implement fast matrix multiplication in OpenCL, it’s 10 pages that introduces new concept little by little: https://cnugteren.github.io/tutorial/pages/page1.html
08:55:21*gokr quit (Ping timeout: 240 seconds)
08:59:37FromGitter<data-man> @mratsim: For CUDA I need some specific version of gcc? In Arch Linux, for some reason, it is suggested to install gcc6. And I do not want to do this. :-)
09:01:22FromGitter<mratsim> yes CUDA 8 requires gcc 5 and CUDA 9 requires gcc 6. Otherwise you can compile cuda kernels with LLVM since a year ago.
09:03:03FromGitter<data-man> Thanks! Does Arraymancer work without external dependencies?
09:04:51FromGitter<mratsim> it requires nimblas, math, nimcuda (though if you don’t pass the -d:cuda flag it won’t tell you to get a Nvidia GPU)
09:04:55*endragor joined #nim
09:06:06*sz0 quit (Quit: Connection closed for inactivity)
09:06:26FromGitter<mratsim> You should use it on nim devel as well. It’s probably 6x-8x faster than 0.17.2 due to useless allocations in the seq code that were fixed by edubart.
09:09:11*endragor quit (Ping timeout: 240 seconds)
09:11:29FromGitter<data-man> Thanks.I plan to add CPUID in stdlib. Which module is most suitable for this? The concurrency/cpuinfo?
09:14:33FromGitter<data-man> Or better to a separate module?
09:18:13*nsf joined #nim
09:21:57*arecaceae quit (Remote host closed the connection)
09:22:16*arecaceae joined #nim
09:24:50skrylarmratsim: sadly haven't done much in ML as of late. been dealing with art and database code
09:41:42*endragor joined #nim
09:44:31Araqdata-man: because they are still in development
09:45:42FromGitter<alehander42> I have an ast represented as a nim variant, and I want to expose `[]` for most nodes (based on a children field), + `.<logical_name>` ⏎ e.g. `Array`: ⏎ ⏎ `node[0]` <=> `node.length` `node[1]` = `node.elementList` ⏎ ... [https://gitter.im/nim-lang/Nim?at=5a1152c6614889d4759c0002]
09:46:07*endragor quit (Ping timeout: 248 seconds)
09:46:33FromGitter<alehander42> my idea is to overload `.` so people can use the node api without much knowledge about the field names, but also use field names for clarity
09:46:40FromGitter<alehander42> (and this way I can also reuse names between branch)
09:53:53skrylarhuh. i guess they are now shipping parts of firefox on rust
09:54:14skrylari didn't mind rust 0.8. some later rusts look like an uglier c++, and it compiles dog slow, but the memory tracing was neat
10:02:05*endragor joined #nim
10:02:06*rauss quit (Read error: Connection reset by peer)
10:02:10*endragor quit (Remote host closed the connection)
10:02:38*endragor joined #nim
10:03:58*yglukhov quit (Remote host closed the connection)
10:04:06*rauss joined #nim
10:06:06*yglukhov joined #nim
10:08:37*yglukhov quit (Remote host closed the connection)
10:09:17FromGitter<alehander42> the nim compiler hangs for me in copyTree o.O
10:10:31FromGitter<alehander42> nah, it just takes a lot of time, is json supposed to support variants
10:19:30FromGitter<krux02> @alehander42 I kind of don't understand your DSL
10:20:48FromGitter<krux02> iirc ``x.foo`` is identical to ``x["foo"]`` is javascript
10:21:29FromGitter<krux02> So I think you could overload the ``.`` operator.
10:21:53FromGitter<krux02> But I am always a bit skeptical about acutally doing it.
10:27:17FromGitter<alehander42> the idea is e.g. you have
10:28:17FromGitter<alehander42> ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a115cc171ad3f8736026f31]
10:30:01FromGitter<alehander42> then you have ⏎ ⏎ ```f = Node(kind: For, children: @[toLabel("child"), toLabel("children"), ..])``` [https://gitter.im/nim-lang/Nim?at=5a115d29e606d60e34e13366]
10:31:03FromGitter<alehander42> and you can access the elements with ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a115d67e606d60e34e13499]
10:31:52FromGitter<alehander42> the mapping is some kind of const array of seq-s ([@["target", "iterable", ..]])
10:33:31FromGitter<alehander42> I know how to implement it, I wonder if there are better ways to achieve something like that ⏎ (notice that with this implementation I can have `label` in many branches too)
10:38:06*sz0 joined #nim
10:39:26FromGitter<mratsim> I’m considering Copy-on-write for Arraymancer tensors, especially because: ⏎ ⏎ 1) I don’t think even with move optimization support you could transform `let a = b[1..2, 1..2] + c` into `let a = b.unsafeSlice(1..2, 1..2) + c` ⏎ 2) Tensors are backed by seq on CPU, and pointers on Cuda (and OpenCL in the future). It’s easier to have seq {.shallow.} by default than giving value semantics to pointers. And
10:39:26FromGitter... it’s definitely better to be consistent across backends. ⏎ ... [https://gitter.im/nim-lang/Nim?at=5a115f5e2837ee5106a06fdc]
10:49:39*marenz__ joined #nim
10:50:49*yglukhov joined #nim
10:51:13*vlad1777d joined #nim
10:51:14skrylarmratsim: that's a lot of copying...
10:55:03FromGitter<mratsim> only slicing b copies here, but often you slice in loops with the loop variable :/
10:56:11*gangstacat quit (Quit: Ĝis!)
11:23:19FromGitter<alehander42> I wrote a macro `genKind(variant, kind)`, would it be useful for the stdlib? it generates an object of `variant` with kind dynamically
11:23:55FromGitter<alehander42> `genKind(MyVariant, M.A) # MyVariant(kind: M.A)`
11:29:22*xkapastel quit (Quit: Connection closed for inactivity)
11:51:33FromGitter<mratsim> seems not far from patty no?
11:52:14*endragor quit (Quit: Leaving...)
11:55:54Araqmratsim: maybe not with moves, but with TR macros?
11:58:14FromGitter<mratsim> I think I would need a TR for reshape, permute, asContiguous, slicing, broadcasting: i.e. all operations that does not touch data but only metadata
11:59:49FromGitter<mratsim> The “move {call}” that is available now works great: ⏎ ⏎ https://github.com/mratsim/Arraymancer/blob/1cfe13c0172dffccbe767bb4aa5a75e8bba1ed67/src/tensor/optim_move.nim#L19
12:00:33FromGitter<mratsim> so that if I do a[1..2].asContiguous.reshape, the asContiguous.reshape are replaced by their unsafe version
12:01:33FromGitter<mratsim> I was thinking that instead I could use shallowCopy by default but with {`var`}, do deep copy.
12:03:08FromGitter<mratsim> i.e. let are shallow, but as soon as a var is involved as input or output everything is deep.
12:08:51Araqthat lets are shallow is actually a bug :P
12:09:13AraqI hope when I fix it I don't introduce performance regressions by using some smart analysis
12:13:22FromGitter<mratsim> Well, hopefully I can use it as well because I have ugly code like this to avoid useless copies on slices: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a117562cc1d527f6b8c04e6]
12:15:03FromGitter<mratsim> vs the much cleaner (and 6x slower currently) ⏎ ⏎ ```code paste, see link``` [https://gitter.im/nim-lang/Nim?at=5a1175c7cc1d527f6b8c05f4]
12:24:46*couven92 joined #nim
12:33:01federico3Araq: regarding #6481 , getNimcacheDir returns a path that does not exists. It seems that 'nim doc' is not generating a nimcache dir or it's generating it elsewhere. Should I create one?
12:47:31*sz0 quit (Quit: Connection closed for inactivity)
13:00:49*endragor joined #nim
13:02:54*SenasOzys__ joined #nim
13:07:38*elrood joined #nim
13:07:45*laas joined #nim
13:11:49*JappleAck joined #nim
13:15:25*couven92 quit (Quit: Disconnecting)
13:16:11*skrylar quit (Remote host closed the connection)
13:25:10laasIs there any way to do a try-catch like thing, but with templates to see if code compiles?
13:25:31*SenasOzys__ quit (Remote host closed the connection)
13:26:42laase.g. `try: syntactically_valid_code except: something_else`
13:29:52*dddddd joined #nim
13:36:02FromGitter<mratsim> You can use assert at compile-time but you cannot use exceptions since they are ref objects
13:37:11FromGitter<mratsim> if you want to run something at compile time wrap it in a static bloc, or use `when foo: doSomething() else: doAnother()`
13:41:52planetis[m]there is when compiles(some code):
13:43:29hohlerdemhmm, do you always get the compiler message "template/generic instantiation from here" when using generic procs within a template?
13:47:58hohlerdehttps://privatebin.tarakis.de/?5c4ff1dc7d4e97c1#WN+iZ6TfDVlANamrjvb/xbyMBhrSo05Nl0vAE1V+VXg=
13:47:59hohlerdebasically I get the compiler message for every "catchError" call.
13:49:59laasthanks planetis!
13:58:13hohlerdeI got rid of the compiler message. i had a deprecation warning for using toLower() in the generic proc. getting rid of that pleased the compiler
14:00:11*gokr joined #nim
14:02:43laasis there a way to err at compile-time?
14:03:09laasI tried raiseAssert in a static block, but it did nothing
14:03:42laasnvm
14:03:44laasit does work
14:05:17hohlerde{.error: "your message".}
14:06:07*hohlerde sent a long message: hohlerde_2017-11-19_14:06:06.txt <https://matrix.org/_matrix/media/v1/download/matrix.org/RqILetpTErSjAGyPLlxzGZQc>
14:07:07*Snircle joined #nim
14:07:25laasah thanks
14:08:00hohlerdelaas: https://nim-lang.org/docs/manual.html#pragmas-error-pragma
14:08:49FromGitter<data-man> @hohlerde: Use toLowerAscii
14:12:46hohlerdedata-man: yes, already changed it and the "template/generic ...." messages went away
14:17:11*vlad1777d quit (Read error: Connection reset by peer)
14:19:58*vlad1777d joined #nim
14:55:55*gangstacat joined #nim
15:01:53*BitPuffin|osx joined #nim
15:02:37FromGitter<alehander42> is there a module that renders valid nim code from nim ast?
15:02:50FromGitter<alehander42> I know about renderer but if that's used in repr(node)
15:03:03FromGitter<alehander42> it seems to render often invalid code
15:03:15*nhywyll joined #nim
15:10:02*noonien joined #nim
15:10:06noonienhello folks!
15:10:09*Arrrr joined #nim
15:10:09*Arrrr quit (Changing host)
15:10:09*Arrrr joined #nim
15:20:44*yglukhov quit (Remote host closed the connection)
15:44:18*miran_ joined #nim
15:44:44ArrrrWhat is nim's killer feature?
15:44:47*miran quit (Ping timeout: 260 seconds)
15:50:20federico3being good at many things instead of having a killer feature
15:51:18*yglukhov joined #nim
15:55:04*JappleAck quit (Quit: Leaving)
15:56:49*yglukhov quit (Ping timeout: 250 seconds)
16:06:28*MJCaley joined #nim
16:10:35*arnetheduck quit (Ping timeout: 240 seconds)
16:14:04*gokr quit (Ping timeout: 268 seconds)
16:19:03*rbrt joined #nim
16:20:32GitDisc<treeform> So, `float` is both a type and a converter yes? `var a: float` and `float(1)`, how can I create a vector type `var a: vec3` that is both a type and a converter `vec3(0,0,0)` but have the name be the same to match how float and built in types work?
16:21:16*MJCaley quit (Quit: MJCaley)
16:30:43*Vladar quit (Quit: Leaving)
16:31:54*astronavt joined #nim
16:33:47*astronavt quit (Remote host closed the connection)
16:33:57*astronavt joined #nim
16:41:35*Sentreen quit (Ping timeout: 248 seconds)
16:43:11*BitPuffin|osx quit (Ping timeout: 250 seconds)
16:48:05*astronavt quit (Remote host closed the connection)
16:48:44*astronavt joined #nim
16:50:03*astronav_ joined #nim
16:50:35ArrrrWhat is vec3, a tupe?
16:53:03*astronavt quit (Ping timeout: 258 seconds)
16:55:01*Sentreen joined #nim
16:56:12*astronav_ quit (Remote host closed the connection)
17:03:20*rbrt quit (Quit: Oíche mhaith)
17:03:48*rbrt joined #nim
17:06:01Arrrr*tuple
17:08:34*astronavt joined #nim
17:10:36*astronavt quit (Remote host closed the connection)
17:22:28*mr_yogurt joined #nim
17:29:48laasIs there something akin to rust's Option?
17:38:37Arrrrhttps://nim-lang.org/docs/options.html
17:39:18ArrrrAlso https://nim-lang.org/docs/theindex.html
17:41:31*ofelas joined #nim
17:44:53*Vladar joined #nim
17:45:45laasthanks
17:45:47laas!
17:45:57laasanother question:
17:46:03laashow do I put \n in a character literal?
17:46:33Arrrr\l
17:47:07Arrrrhttps://nim-lang.org/docs/manual.html#lexical-analysis-string-literals
17:47:09laaswhat's the difference?
17:47:14laasthanks
17:47:43ArrrrFor some technical reason \n is not allowed in nim as a literal char
17:50:24*SenasOzys joined #nim
17:52:41planetis[m]i think /n is two characters depending on the os
17:56:20planetis[m]do you have windows; can you try echo "\n" == "\c\l"
17:57:20dom96Yes, \n is platform dependent.
18:01:35*Arrrr quit (Ping timeout: 240 seconds)
18:02:12laasah ok
18:02:16laasthat's pretty nice
18:03:19*astronavt joined #nim
18:05:52*yglukhov joined #nim
18:06:54*Arrrr joined #nim
18:09:04*astronavt quit (Remote host closed the connection)
18:10:01*yglukhov quit (Ping timeout: 240 seconds)
18:21:09*astronavt joined #nim
18:24:45GitDisc<treeform> Arrrr, just a normal type https://gist.github.com/treeform/4f3811cb0e32284f65059781366bd7bd
18:25:19GitDisc<treeform> I am just asking is there a way to have `newV3` have name of `v3` some how?
18:25:25GitDisc<treeform> So it matches built in types.
18:25:30*mr_yogurt quit (Disconnected by services)
18:25:31*yglukhov joined #nim
18:27:51*Arrrr quit (Ping timeout: 240 seconds)
18:28:30*MJCaley joined #nim
18:35:29*Arrrr joined #nim
18:35:59ArrrrYes, you can do `v3(x: 1, y: 2, z: 3)`
18:36:27GitDisc<treeform> yes, but I don't want to type all of that out every time.
18:36:45GitDisc<treeform> there is no way I can do v3(1, 2, 3)?
18:37:02GitDisc<treeform> like I can do with float(1) and int(2.3)?
18:37:14ArrrrNo. However, you could define v3 as a tuple
18:37:44GitDisc<treeform> then overload x= and x functions?
18:37:55GitDisc<treeform> for getting [0] element?
18:38:13ArrrrNot necessary, look at this example
18:40:17Arrrrhttps://play.nim-lang.org/?gist=b9789a1afdec10f8af11d443d6f83c29
18:40:42ArrrrNim is very anal with types, i thought it would not be necessary to add '.0' to detect a float
18:41:54GitDisc<treeform> hmm so I would use (1,2,3) + (1,2,3) without the v3 event?
18:42:06GitDisc<treeform> v3 name*
18:42:21ArrrrYeah, you could do that
18:43:04GitDisc<treeform> cool ill try this
18:43:26GitDisc<treeform> it should work
18:43:28ArrrrYou could also rename your type to 'V3' and call the proc 'v3'
18:43:37Arrrror 'Vector3'
18:43:43GitDisc<treeform> yes I did that at first
18:44:10GitDisc<treeform> in other langauges I called my types v3 and am used to it.
18:44:15GitDisc<treeform> but they had constructors
18:44:29GitDisc<treeform> v3 type and v3() constructor
18:44:41GitDisc<treeform> but maybe tuples are event better
18:44:47GitDisc<treeform> my code is littered with v3s
18:46:34GitDisc<treeform> I am working a font rasterizer, inspired by this dude: https://github.com/google/font-rs
18:47:02GitDisc<treeform> I don't have all of my curves right yet: https://dl2.pushbulletusercontent.com/tKAD614SiRICv5vcc8U1zk29m8jHgYjO/image.png
18:48:26FromGitter<Varriount> treeform: I think the O's look pretty
18:48:53GitDisc<treeform> You should see my heart shape
18:49:14GitDisc<treeform> oh I did not save it
19:14:05*nhywyll quit (Ping timeout: 240 seconds)
19:15:46*rbrt quit (Quit: Oíche mhaith)
19:16:48*mr_yogurt joined #nim
19:18:40miran_there is no `toU64`, like there are `toU8`, `toU16` and `toU32`?
19:19:18miran_or: how do i multiply uint64 and int? :)
19:20:02mr_yogurtare there any plans for user-addable binary operators that don't use punctuation symbols? maybe declared with something like proc ` opname `(...): ... = ... with spaces before & after since you'd have to add space to use them like with the builtins
19:20:28Araqmr_yogurt: never gonna happen, makes parsing dependent on a symbol table
19:20:50laaswould conflict with commands maybe?
19:20:54Araqmiran_: use a type conversion
19:22:19laasis there any way to control an iterator's state manually?
19:22:25miran_Araq: thanks, works
19:22:50mr_yogurtright, since 'a b c' could be ambiguous without knowing about what a, b, c are
19:22:57Araqyes
19:27:26*nhywyll joined #nim
19:29:38Araqmr_yogurt: Fortran uses .op. iirc
19:29:48Araqbut it's pretty ugly
19:30:01*xkapastel joined #nim
19:30:08Araqmaybe some other lexical rule can help
19:30:24*vlad1777d quit (Remote host closed the connection)
19:31:43mr_yogurtwell with the UFCS i guess a .b c works
19:31:54Araqtrue
19:33:30*rbrt joined #nim
19:36:29FromGitter<alehander42> Araq: can renderer.nim be made to produce valid code easily?
19:37:11Araqdepends on what you mean by it
19:37:24Araqusually it emits valid Nim code
19:37:39FromGitter<alehander42> not always, I have to generate Nim code for something
19:37:49FromGitter<alehander42> and I wonder should I emit NimNodes and repr them
19:37:54FromGitter<alehander42> or directly generate the code
19:38:15AraqOrmin uses staticWrite and repr and that's what you should use too
19:38:47FromGitter<alehander42> most of the macros I write generate invalid code in repr
19:40:02*PMunch joined #nim
19:40:46FromGitter<alehander42> but if I am careful with the generation, I think it will work ok
19:53:50*yglukhov quit (Remote host closed the connection)
20:01:53*Trustable joined #nim
20:08:37*gokr joined #nim
20:09:48*NimBot joined #nim
20:17:07*rbrt quit (Quit: Oíche mhaith)
20:18:04*rbrt joined #nim
20:40:26*Ven joined #nim
20:40:51*Ven is now known as Guest57431
20:44:48euantorI know that concepts are a work in progress, but is this intended? https://play.nim-lang.org/?gist=b2a465a953d4d49deb22160516b5d4e6
20:45:19euantorIf you remove the `posix` import, the result is `true, true`. With `import posix`, you get `false, true`
20:45:27euantorSeems like somehting related to name resolution?
20:47:13*rbrt quit (Quit: Oíche mhaith)
20:50:49*Arrrr quit (Read error: Connection reset by peer)
20:54:51*rbrt joined #nim
21:10:56*Guest57431 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
21:11:02Araqeuantor: lol, wtf
21:12:03euantorYeah, an interesting one imo
21:12:33euantorTook me a bit of playing to find the problem. Just playing with Concepts to see what they can do
21:13:16*Ven joined #nim
21:13:40*Ven is now known as Guest38788
21:15:07*ofelas quit (Quit: shutdown -h now)
21:16:34*JappleAck joined #nim
21:16:46*nsf quit (Quit: WeeChat 1.9.1)
21:21:00AraqI think I know why it fails though and it's just a bug
21:21:30euantorShall I report it on GitHub?
21:21:39Araqyes please
21:22:12euantorNo problem, will do so in a while
21:25:49*Vladar quit (Quit: Leaving)
21:25:50*sendell joined #nim
21:27:18Araqimport posix except read makes it work
21:27:28sendellhey guys :) Is vscode pluggin's autocompletion suposed to work with latest nim & VsCode ?
21:27:54sendellnot working on both windows and linux for me :/
21:31:17*jsgrant joined #nim
21:39:35*miran_ quit (Quit: Konversation terminated!)
21:41:44Araqsendell, too bad but it is supposed to work
21:46:23*rbrt quit (Quit: Oíche mhaith)
21:46:24sendellglad to hear that :) i'll keep searching then
21:47:04*rbrt joined #nim
21:47:07*MJCaley quit (Quit: MJCaley)
21:48:34*Jesin quit (Quit: Leaving)
21:49:51*rbrt quit (Client Quit)
21:51:19*rbrt joined #nim
21:53:54federico3Araq: adding a new parser error to docgen.compilerMsgHandler for #6481 looks a bit messy: this is a doc "rendering" error rather than compiler/parser. Are you OK with it?
22:03:12*bluenote joined #nim
22:04:57bluenoteHello everyone! Quick question: If I have a NimNode from getType within a macro, how can I convert this NimNode into the corresponding typedesc?
22:09:28*sendell quit (Remote host closed the connection)
22:09:33*Guest38788 quit (Quit: My MacBook has gone to sleep. ZZZzzz…)
22:10:47*vlad1777d joined #nim
22:15:50*PMunch quit (Quit: leaving)
22:17:31*claudiuinberlin quit (Quit: Textual IRC Client: www.textualapp.com)
22:18:41*Trustable quit (Remote host closed the connection)
22:24:11FromGitter<mratsim> For use in a macro or in a non-completion proc?
22:26:58bluenoteyes, for use in a macro at compile time. For example, I have mymacro(x: typed) so I can get x.getType(). But now I want to call a function f(T: typedesc) at compile time in the macro. So I need a function converting the NimNode to typedesc
22:27:09*rbrt quit (Quit: Oíche mhaith)
22:27:37*vivus joined #nim
22:28:24FromGitter<mratsim> Mmh would proc f(T: NimNode) work instead?
22:29:49bluenoteit would compile, but what I'm trying to achieve requires calling a function with a typedesc argument
22:31:16FromGitter<mratsim> Maybe use quote do syntax
22:32:03*astronavt quit (Remote host closed the connection)
22:32:06FromGitter<mratsim> Or newCall to construct the call in the macro
22:32:15*rbrt joined #nim
22:32:17*astronavt joined #nim
22:32:36*Jesin joined #nim
22:32:53*astronavt quit (Remote host closed the connection)
22:33:04FromGitter<mratsim> I'm not aware of a way to transform a typedesc NimNode to typedesc
22:33:27bluenoteyes exaclt, I tried newCall, but for some reason the argument stays of type NimNode and is not allowed as a typedesc
22:33:28FromGitter<mratsim> (but I'm a beginner at Nim metaprogramming)
22:34:19FromGitter<mratsim> Quote do / getAST maybe then
22:36:40*astronavt joined #nim
22:36:53*astronavt quit (Remote host closed the connection)
22:40:01*nhywyll quit (Quit: nhywyll)
22:52:41bluenotehm, that all gives me NimNodes. Tricky to escape from the NimNode world, maybe this is not even supposed to be possible?
22:52:47*bozaloshtsh quit (Ping timeout: 248 seconds)
22:53:44*bozaloshtsh joined #nim
22:53:44*bozaloshtsh quit (Changing host)
22:53:44*bozaloshtsh joined #nim
23:20:16*elrood quit (Quit: Leaving)
23:46:24*bluenote quit (Ping timeout: 260 seconds)
23:46:40*gokr quit (Ping timeout: 248 seconds)
23:55:09*mr_yogurt quit (Ping timeout: 260 seconds)