🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). Log available at irclogs.raku.org/raku/live.html . If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 6 September 2022.
00:02 apac left 00:08 LainIwakura left 00:11 arkiuat left 00:21 arkiuat joined 00:26 arkiuat left 00:30 Sgeo joined 00:51 arkiuat joined 00:56 arkiuat left 01:15 arkiuat joined 01:28 hulk joined 01:29 kylese left 01:48 Aedil joined 02:15 hulk left, kylese joined 02:50 patrickb left, patrickb joined 03:18 deoac left 03:36 arkiuat left 03:45 kylese left 03:49 arkiuat joined 03:53 arkiuat left 04:13 arkiuat joined 04:18 arkiuat left 04:27 arkiuat joined 04:35 arkiuat left 04:48 arkiuat joined 04:54 arkiuat left 05:07 arkiuat joined 05:11 arkiuat left 05:18 arkiuat joined 05:23 arkiuat left 05:29 Sgeo left 05:35 arkiuat joined 05:39 arkiuat left 05:46 LainIwakura joined 06:05 arkiuat joined 06:09 arkiuat left 06:28 kylese joined 06:34 arkiuat joined 06:35 silug left 06:40 arkiuat left 06:48 silug joined 07:10 LainIwakura left 07:13 arkiuat joined 07:17 arkiuat left 07:19 arkiuat joined 07:23 arkiuat left 07:25 LainIwakura joined 07:35 lichtkind joined 07:52 arkiuat joined 07:56 arkiuat left 08:10 arkiuat joined 08:11 dakkar joined 08:15 arkiuat left 08:24 Manifest0 joined 08:25 arkiuat joined
disbot4 <librasteve> melezhik: please can you “weekly” this (i can’t do it from discord) 08:26
<melezhik.> weekly: news.ycombinator.com/item?id=44726955 - Application less containers 08:27
08:33 arkiuat left
lizmat weekly: news.ycombinator.com/item?id=44726955 08:37
notable6 lizmat, Noted! (weekly)
guifa two new realizations in working with nativecall 08:47
1) we don't actually have a guide on how to do callbacks (there's just a stackoverflow post which feels less than optimal) 08:48
2) it's possible to have a sub and a method with the same name in a class. I know why, but it still feels weird haha, but makes my life a lot easier tbh 08:49
09:01 arkiuat joined
Voldenet there is some guide on callbacks in docs.raku.org/language/nativecall (see Function arguments and Function pointers) but it isn't really exhaustive 09:04
09:05 arkiuat left
guifa Voldenet: yeah, but it's very brief. I think it could use some "what do things look like on the C side" / a toy example. I thikn I can write that up this week 09:13
maybe too I can turn this little adventure into a nice advent calendar post too 09:14
09:16 apac joined 09:21 arkiuat joined 09:26 arkiuat left 09:39 kaskal left 09:44 kaskal joined, inspork left 09:46 inspork joined 09:53 arkiuat joined 09:58 arkiuat left 10:07 melezhik joined
melezhik asciinema on managing Alpine Linux with Sparrow - asciinema.org/a/730670 10:07
tellable6 2025-07-27T03:19:47Z #raku <wayland76> melezhik Yeah, I didn't ask for the comparison to RakuDist. Feel free to remove that part, and replace it with a comparison to chef/ansible/cfengine instead :) .
2025-07-27T03:21:00Z #raku <wayland76> melezhik And, it sounds like the questions should be re-asked, but with Sparky as part of the mix. My point is, these are the kinds of things that I think people visiting the Sparrow site will want to know.
melezhik added to HN as well - news.ycombinator.com/item?id=44732514 10:08
10:14 melezhik left, arkiuat joined 10:19 arkiuat left 10:24 Aedil left 10:29 Aedil joined 10:39 arkiuat joined 10:44 arkiuat left 11:17 arkiuat joined 11:30 arkiuat left 11:32 LainIwakura left 11:38 arkiuat joined 11:45 arkiuat left 11:50 LainIwakura joined 11:53 apac left 12:10 arkiuat joined
disbot4 <antononcube> @guifa I have been looking into packages of Itsuki Toyota (titsuki) and Brian Duggan (bduggan) for using Raku's Native Call. 13:01
<antononcube> I created two very important for me native-call-based packages: "Math::SparseMatrix::Native" and "Math::DistanceFunctions::Edit". 13:03
<antononcube> ... more or less, following titsuki's "pattern" of Native Call utilization. 13:04
13:20 LainIwakura left 13:22 guifa left 13:27 LainIwakura joined 13:46 om joined 13:48 apac joined 13:49 LainIwakura left 13:52 om82 joined 13:53 om82 left 13:54 om left 14:04 LainIwakura joined 15:02 cpli joined
cpli heyya, thought of writing some cute offline dsp programs, and noticed Buf doesn't allow for writing num16... 15:02
actually the entire language seems to lack num16 is that deliberate, they appear quite often in dsp q-q 15:03
though i guess ffmpeg doesn't even have support for f16le—at most providing support for f32le—so maybe i should just swallow the buffer sizes 15:06
15:22 librasteve_ joined
disbot4 <librasteve> cpli: afaik raku has not yet implemented num16, num8 and so on … actually the language is fine for it (although likely some of the alternates for very short floats will need some names) as the notion of num32, num16 implies, it’s actually a not implemented aspect I think … all the hooks such as NativeCall and coercion, even infectiousness are there … so would be a great module project 15:31
15:32 LainIwakura left
disbot4 <antononcube> For some machine learning computationsnum16 and/or num8 are useful, because of compactness and speed when using dense vectors and matrices. 15:36
<antononcube> I am not sure what would be the priority to introduce those types, though. I would say "low", except, if implementing them is a low hanging fruit. 15:40
cpli librasteve, would it actually be possible to simply add those features as a module, i would have thought native data types are too intrinsic to be implemented outside the interpreter 15:44
antononcube, honestly currently listening to `ffmpeg -i "$m" -f s8 - | ffplay -ch_layout stereo -` and it isn't too offputting, so i'll just play around with signed bytes for a bit 15:50
Voldenet maybe I'm crazy but it seems like a very low hanging fruit 15:51
it could be enough to add it to typemap in NativeCall.rakumod 15:52
and NativeHOW.nqp
tbrowder .tell zorfnorf go to channel #raku-beginner and we can help better there 15:55
tellable6 tbrowder, I'll pass your message to zorfnorf
Voldenet yeah, nevermind, it's not enough
> github.com/MoarVM/MoarVM/blob/b1c7...num.c#L107 15:58
> github.com/MoarVM/MoarVM/blob/b1c7...P6num.h#L6
it seems like low hanging fruit still, but it's not trivial 15:59
16:15 LainIwakura joined 16:17 apac left 16:39 dakkar left
disbot4 <librasteve> cpli: yeah you are probably right, I as thinking that most of the stuff is there from a module - but you probably need to hook the compiler and actually add those native types as Voldenet mentions [well you could write a slang to make up your own types in a module but I'm ignoring that] 16:46
lizmat Voldenet: could you make a MoarVM issue for that ? 16:49
16:52 LainIwakura left 17:03 LainIwakura joined
disbot4 <librasteve> who is the "owner" of the channel controls on the Discord Raku server? please 17:05
<librasteve> [ie the admin] 17:06
lizmat disbot4: help 17:10
meh
disbot4 <antononcube> 🙂 17:11
<nahita3882> Rog 17:18
<librasteve> tx 17:26
17:27 kybr joined
disbot4 <librasteve> hi @.landyacht. i have a couple of change requests for the raku Discord server (eg remove/archive some channels to keep it lively and reduce confusion) … i suggest to make a raku discussion issue on github to encourage others to comment on my ideas and then to ask you to make the changes according to the consensus… is that process ok by you 17:30
<.landyacht.> Sure, that works for me 17:33
<.landyacht.> Depending on the results I may apply some level of discretion, but I will try to follow the consensus as much as possible 17:34
<librasteve> cool - thanks!
<librasteve> will share here so that you can point our when we suggest the impossible ;-) 17:35
17:35 LainIwakura left
arkiuat I'm using NeoVim on MacOS more or less out-of-the-box, and it's annoying me that comment text is the same color (light blue) as variable names. Looking at the vim.raku from github.com/Raku/vim-raku, I find that the syntax file makes no direct reference to any colors, but presumably hooks up with something which does and which can be edited. Any tips? 17:36
disbot4 <librasteve> (ie share the issue url)
17:39 LainIwakura joined 17:42 LainIwakura left
arkiuat n/m I found the info in neovim.io/doc/user/syntax.html 17:42
17:49 undermine left
arkiuat wow, I just changed three color settings and it's SO much better :D 17:51
lizmat nice reference 18:29
19:13 stanrifkin joined 19:14 arkiuat left 19:19 arkiuat joined 19:23 arkiuat left 19:25 apac joined 19:38 Aedil left 19:53 arkiuat joined 19:57 arkiuat left
Voldenet lizmat: I will, but honestly I have no idea how would one express num8 in C, there's only C23 proposal for num16 20:18
lizmat but wasn't it the point that some LLM software uses some representation of those tiny nums? wouldn't that be a good start? 20:19
anyway, my point was to make an issue, so it doesn't fall through the cracks
Voldenet yeah, making it right now
disbot4 <librasteve> www.perplexity.ai/search/i-am-writ....bZkdhM2QA 20:25
20:25 arkiuat joined
disbot4 <librasteve> looks like there is a need to distinguish between vector types (eg. CUDA shader float4) and regular (ie scalar) types ... I expect with Raku that the vectors are out of scope in the absence of a big extension (think PDL) 20:28
20:28 inspork_ joined 20:30 inspork left, inspork_ is now known as inspork
Voldenet I hope I didn't miss anything discussed before github.com/MoarVM/MoarVM/issues/1951 20:37
21:30 arkiuat left 21:42 arkiuat joined 21:47 arkiuat left 22:09 lichtkind left 22:13 stanrifkin left 22:14 arkiuat joined 22:18 apac left 22:19 arkiuat left 22:20 guifa joined 22:35 Sgeo joined 22:38 arkiuat joined 22:43 arkiuat left 22:52 arkiuat joined 23:08 librasteve_ left
arkiuat one of the things I wish I had learned earlier than I did was that .keys and .kv can be applied to arrays to get the indices. It certainly makes it easier to switch back and forth between array and hash when trying different things out 23:13
m: my IntStr $l = 5 23:25
camelia Type check failed in assignment to $l; expected IntStr but got Int (5)
in block <unit> at <tmp> line 1
arkiuat m: my IntStr $l = "5"
camelia Type check failed in assignment to $l; expected IntStr but got Str ("5")
in block <unit> at <tmp> line 1
arkiuat Is it just not possible to assign to IntStr or other Allomorphs?
oh wait, there's an IntStr.new method but it doesn't seem to be documented? 23:26
oh wait, it is documented. I was looking in Allomorph's docs and not IntStr's, sorry 23:27
so that answers my earlier question 23:28

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!