01:20 kylese left, hulk joined 02:15 hulk left, kylese joined 03:22 lichtkind__ joined 03:24 lichtkind_ left 03:44 runesicle joined
Geth docker: f9a08f038e | AntonOks++ | 6 files
Bump to 2026.03 [skip workflow]
 03:48
03:57 runesicle left 04:12 Aedil joined 04:16 Aedil left 04:19 Aedil joined 07:06 sibl joined 07:07 Sgeo left 07:47 annamalai left 07:48 sibl left, annamalai joined, sibl joined 08:03 dakkar joined 08:04 sibl left, sibl joined 08:12 Aedil left
disbot2 <simon_sibl> I do a lot of silly action with csv file, I have a 10 lines raku file that parse 2 csv file and basically merge them with a simple condition, isnt there a cli tool that would do that ? or the 10 lines script is probably more flexible ? 09:50
timo @simon_sibl maybe this does what you want? miller.readthedocs.io/en/latest/ 09:58
disbot2 <simon_sibl> wow nice, thank you ! indeed thats very handy cli for that 10:01
timo it looks very very powerful, with everything that implies :) 10:02
10:06 Aedil joined
timo apparently I used it with this script before to analyze NFAs run during core setting compilation: mlr --icsv --ocsvlite head -n 10000000 then filter '$eos == 0x32aea9' then put -s entry=0 '$len = $offset - $orig_offset + 1; $entry = @entry; @entry += 1' then cut -f nfa,entry,orig_offset,offset,len,nfa then repeat -f len then put -s curpos=-1 'if (@curpos == -1) { $pos = fmtnum($orig_offset, "%d"); 10:13
@curpos = $orig_offset + 1 } else { $pos = fmtnum(@curpos, "%d"); if (@curpos == $offset) { @curpos = -1; } else { @curpos += 1; }; }' then cut -f pos then uniq -c -f pos then sort -n -f count core_setting_nfa_stats.csv.zst | zstd -v -17 > what_position_how_often.csv.zst
10:17 annamalai left, annamalai joined 10:26 sorenson left 10:27 sorenson joined
Geth CCR/main: f2704a9113 | (Elizabeth Mattijsen)++ | 39 files
Remaster pmichaud's blog bposts on pmthium
 10:53
lizmat wonders if Crag is able to convert gas prices in $/gallon to ā‚¬/liter 11:06
and vice-versa 11:07
but is too lazy to figure it out herself :-)
Geth CCR/main: c277384487 | (Elizabeth Mattijsen)++ | Remaster/Patrick Michaud/README.md
Fix index rendering issue
 11:09
timo lizmat: it's supposed to be able to do that, yeah 12:49
lizmat hmmm... maybe we should run a Crag bot on the channel :-)
13:05 sibl left, sibl joined 14:02 sp1983 joined
sp1983 tell: El_Che I've tested your aarch64 rpm - sparky.sparrowhub.io/report/test-u...ault/28545 looks good to me 14:04
tellable6 sp1983, I'll pass your message to El_Che
14:08 sp1983 left
disbot2 <librasteve> yes 14:10
<antononcube> Can we have a non-vi-users channel first? 14:11
<antononcube> @librasteve This reminds me -- did you make crag available within Raku sessions? 14:14
<antononcube> I remember we discussed it, but I do not think I filed a GitHub issue for that.
<librasteve> ^<20 USD / gallon> in 'EUR / litre' 14:15
<librasteve> thanks for the reminder ā€¦ sadly I have not done that (yet) 14:16
<librasteve> one way of thinking about crag is that itā€™s a pioneer for a new Raku slang ā€¦ but I havenā€™t really got my head around that 14:17
<librasteve> in the example above. maybe the gland can eliminate the need for ^<> 14:19
<antononcube> Here the CLI of the Raku package "WWW::WolframAlpha" is used: > wolfram-alpha convert 20 gallons to liters 75.71 L (liters)
<antononcube> You can see the "official" result here: www.wolframalpha.com/input?i=conve...+to+liters
14:21 sibl left
disbot2 <antononcube> Ha! 20 gallons is "ā‰ˆ 1.1 Ɨ average volume of the human body ( ā‰ˆ 66.4 L )". I am not sure is that an average US human body or a "world-wide" human body. 14:23
<librasteve> ok I have a short alternative- will try to release that later today 14:26
<librasteve> short term
<antononcube> Unfortunately, Wolfram|Alpha cannot give a proper answer to: > How many bananas would you need to power a house? 14:30
<librasteve> $be = ?^<banana energy in kcal> 14:36
<librasteve> $hp = ?^<average house power in kW> 14:37
<librasteve> $hp / ($be / ^<1 day>) #348
<librasteve> please can someone make @simon_sibl an editor in the Whaterables wiki (in the raku GH org)? 14:40
<antononcube> I think the house powering with bananas is based on their radioactivity (having potassium.) 14:41
<librasteve> (?^<banana mass in kg> * c**2) / ^<1 day> 14:44
<librasteve> the answer is 83 million houses for a day 14:46
<antononcube> what-if.xkcd.com/158/ 14:48
15:01 orangebot joined 15:22 soverysour joined, soverysour left, soverysour joined 15:25 Aedil left 15:26 soverysour left 15:29 soverysour joined, soverysour left, soverysour joined 15:31 Aedil joined
lizmat People complaining about the price of gas in the US: 15:38
% crag '^<2.5 EUR / litre> in "USD / gallon"'
13.12USD / gallon
that's what we pay in the NL for gas
ugexe yeah but how much do you pay for insulin? 15:45
we need cheap gas prices to drive to canada and mexico for healthcare 15:47
lizmat I guess that makes sense... if you're a diabetic with a driver's license :-) 15:49
which I understand is quite a big group in the US :-(
disbot2 <librasteve> lizmat \o/ 16:03
<librasteve> I guess you want US gallons ā€¦ ā€œus-gallonā€ ā€¦ Crag picks the default depending on your locale 16:09
<librasteve> are you using en_uk (if you donā€™t mind me asking)? ā€¦ guess I should add a warn during install 16:11
[Coke] My once-a-day insulin is pretty expensive, and I have what is considered good insurance. 16:20
... though it did go down a bit under the last administration and has held there. 520USD for 3 months. (top of year, I pay all of that - as I reach the deductible & total out of pocket, the cost per quarter to me is reduced) 16:26
as little as 2 years ago, that was more like 1300USD
16:31 dakkar left
disbot2 <librasteve> yeah the NHS has its problems ... but its easy to forget the upside 16:34
16:44 soverysour left
[Coke] m: put 2026-1953 16:52
camelia 73
lizmat % crag '^<2.5 EUR / litre> in "USD / us-gallon"' 17:03
10.93USD / us-gallon
ok, so gas is less than 3x as expensive than in the US
disbot2 <librasteve> I am reviewing the way Physics::Unit checks locale and sets up 'us-' or 'imp-' as the default btw 17:06
[Coke] huh. ziptbl.com is nifty: shares data directly in the URL: ziptbl.com/#d=eNoFwbcBACAMwLA9t3jA...WN8wFxQAZX 17:07
lizmat interesting... also from a surveillance point of view 17:08
as the URL is considered (trappable) metadata
[Coke] don't have to store it on a server, anyone with the link can view the data... Wouldn't use it for anything I needed secure, but perhaps a nice way to dump a bunch of json data. 17:09
disbot2 <librasteve> that's wild (the way they encode the data)
[Coke] tries a BUNCH of json data...
ok, too big to shove all the blin results into. :) (it works, but the resulting url is 5000140 bytes. 17:11
17:11 soverysour joined, soverysour left, soverysour joined
disbot2 <librasteve> lol 17:12
17:12 david7832 joined
[Coke] down from 40755394 on the original JSON 17:12
lizmat and if you zip the blin results? 17:13
just OOC
disbot2 <librasteve> I figure that only GB (maybe australia, canada, india?) wants imperial gallons
<librasteve> - any feedback welcome 17:14
[Coke] gzip is 3886990
I do not wish imperial anything (except pints when drinking)
17:14 runesicle joined
lizmat I guess that's what you get for using something like base64 instead 17:15
[Coke] bz2 better at 3056874
lizmat zip and then express in base62
*64
[Coke] Yup. Nifty but probably just a toy.
david7832 seems surveillance isn't even an issue: the data in the URL is after the # , so it doesn't even become part of the request (see link "How it works") 17:16
tellable6 2026-03-29T14:09:04Z #raku <librasteve_> david7832: my understanding is that the current push to RakuAST will open the door to revisit the whole 007 macro experiments
[Coke] Might be nifty for something like the Text::CSV results.
david7832: at time of use, sure. they have to get the URL somehow
lizmat on further reading, looks like the data never makes it over the line... interesting 17:17
disbot2 <librasteve> if only they could use unicode props in the fragment 17:22
<librasteve> (more chars = better encoding), right?
lucs My rakubrew installation appears to have made `rakubrew` a shell function, not an executable found in the PATH. 17:37
So, not sure why, but when in Raku I do something like "shell rakubrew" or "run rakubrew", I get a "command not found" or a "spawned command 'rakubrew home' exited unsuccessfully".
Any way to have it be an executable found in the path instead?
lizmat invokes patrickb 17:38
lucs (Er, copy/paste error: just "... command 'rakubrew'", not "... command 'rakubrew home'") 17:39
patrickb lucs: rakubrew mode env
lucs Hmm... My MODE file does say 'env'... 17:40
patrickb but I'm pretty sure this won't change how rakubrew is invoked, only how raku and scripts are invoked 17:41
lucs And 'rakubrew mode' does report 'env'. 17:42
patrickb yeah then it already is in env mode.
you could change the .bashrc invocation to just be a export PATH=... thing 17:43
but possibly rakubrew will complain that the hook is Missing 17:44
lucs Not sure how I decided to use this, but I currently have `eval "$($RAKUBREW_HOME/bin/rakubrew init Zsh)"` (I'm in zsh, eh.) 17:45
I'll try to figure out what init does. 17:46
Only 49,842 lines in the rakubrew scrip t :~( 17:47
*script 17:48
disbot2 <librasteve> lizmat: your local (unless it is 'en_GB') will now select us-gallons by default (need to reinstall Physics::Unit and Physics::Measure) - sorry 17:53
patrickb lucs better look at the repo. That script is a Perl packed version. 17:55
disbot2 <librasteve> may take a few minutes for zef to index them
lucs patrickb: Yeah, just realized, thanks. 17:56
patrickb github.com/Raku/App-Rakubrew/blob/...ell/Zsh.pm
in short, the shell hook allows rakubrew to call shell functions in the users shell. e.g. change PATH. 17:57
lucs Is it correct to say then that `eval "$($RAKUBREW_HOME/bin/rakubrew init Zsh)"` must be run in any new shell session? (I think that may be my problem.) 17:58
patrickb that rakubrew shell function retrieves shell code to run in the users shell 17:59
github.com/Raku/App-Rakubrew/blob/...rew.pm#L77
that's the core of the messy stuff.
looking at the comment a few lines down seem to indicate you can run without the shell hook if you switch to shim mode
lucs: I believe so, yes. 18:00
that eval creates that shell function
github.com/Raku/App-Rakubrew/blob/...Zsh.pm#L75 18:01
that's the hook itself
afk for a while
lucs Thanks for the pointers.
patrickb but do shoot me any questions you have. I'll answer when I have a free moment.
timo fwiw, the fragment part of an URL isn't sent to the server by the browser, only if there's some javascript on the page that decides to send it 18:07
there's many pages like zerobin for example that put something there that should not go to the server for this reason
this is also what the new kernel panic screen on linux has, a huge qr code that links to a static web application that decodes and displays the data for you 18:08
lizmat feels like we should have something like that for Raku as well 18:10
disbot2 <librasteve> @antononcube ... thanks for suggesting to make App::Crag callable from within a Raku script ... I have had a look at this 18:18
<antononcube> I wanted to do an examples and comparisons with unit calculations. In a Jupyter notebook. 18:19
<librasteve> problem is that App::Crag is a layer designed to be called from the command line and/or as a REPL and tbh to make it just a plain module is gonna be a megahack ... I cannot work out a way to do it that is not perverse
<librasteve> otoh, we can easily go my $res = qqx(crag '0.1+0.2'); 18:21
timo not sure what exactly the "something like that" should be
the kernel panic qr code thing is because when your kernel paniced you can't copy&paste or save to disk your output
disbot2 <librasteve> is that something you can add as a Jupyter::Chatbook "shortcut" 18:22
SmokeMachine lizmat: I didnā€™t know about roles not having their variables (my English is missing while I write this sentence). I tried using `has`, but when I tried accessing %types from inside the role, it errored I was trying accessing %!typesā€¦ :( but if I declare my %types before the role, that worksā€¦ 18:23
lizmat: I didnā€™t know about roles not having their variables (my English is missing while I write this sentence). I tried using `has`, but when I tried accessing %types from inside the role, it errored I was trying accessing %!typesā€¦ :( but if I declare my %types before the role, that worksā€¦ 18:25
lizmat SmokeMachine: have considered parameterizing the role with the prefix ?
18:27 sibl joined
SmokeMachine lizmat: that works as well, thanks! glot.io/snippets/hh8ma9c73o 18:27
lizmat no, not like that :-) 18:28
that will just give each concretization its own %types
SmokeMachine But I still think that the original code shouldnā€™t break on Grammar, and try to use %!types when I declared %types is confusing. 18:29
lizmat: So, how did you mean?
lizmat: and yes, you are correct: glot.io/snippets/hh8ma9c73o 18:31
lizmat gist.github.com/lizmat/d0cc2c3195d...05e4623b4b
with possibly an error message if there was already a class with a given prefix ? 18:32
note that the gist contains a race condition on the binding to the hash 18:36
but with a very small chance, only if two thread decide at the same time to consume the tole 18:37
*treads
*threads
grrr
SmokeMachine That was my first tryā€¦ (github.com/FCO/IdClass/commit/e501...debd7L168) but I wanted it to be possible to be dynamic and it was conflicting with the method optionā€¦ then I removed itā€¦ 18:39
18:46 soverysour left 18:47 soverysour joined 19:28 tejr left, tejr joined
disbot2 <antononcube> Yes, there can be a special "Jupyter::Chatbook" magic cell that invokes "crag". For example: %% crag (?^<banana mass in kg> * c**2) / ^<1 day> 19:35
<antononcube> This is similar to the crag-slang idea. 19:36
<librasteve> that seems like a good solution - tx! 19:48
19:48 sibl left
disbot2 <librasteve> I will bring the crag DSL up my priority list 19:49
19:51 silug left 19:54 silug joined 20:03 Aedil left
lucs patrickb: I tried simplifying my rather complicated shell spawning setup, but have the same problem. 20:06
patrickb: Tell me, does this work for you?: `raku -e 'shell rakubrew'`
Um, and anyone else having set up rakubrew, does that work for you? 20:07
librasteve_ www.irccloud.com/pastebin/FSb8CxrN 20:09
TLDR: nope
timo i think it was meant to be "rakubrew" rather than just rakubrew 20:11
librasteve_ www.irccloud.com/pastebin/MrioQMZx 20:13
TLDR: kinda
www.irccloud.com/pastebin/yKYgdsMN 20:14
seems that `rakubrew init` is not fixing that ā€¦ otherwise rakubrew is working fine for me (in macOS) 20:15
[Coke] shell('rakubrew') would work if it weren't an alias.
try 'which rakubrew' in your shell and see what it emits.
librasteve_ ^^ www.irccloud.com/pastebin/8hSejdxQ 20:16
lucs [Coke]: `which rakubrew` shows me its function definition, similar to that librasteve_ paste. 20:18
[Coke] Yup. So that isn't going to work in a shell() call 20:19
I'd expect to look just like: /bin/sh -c rakubrew
sorry, function, not alias.
patrickb lucs: does rakubrew run when you call it with it's full path? (i.e. not the shell function)
You can also try passing the same args that the shell function would pass. 20:20
lucs patrickb: It complains like librasteve_'s paste: www.irccloud.com/pastebin/MrioQMZx
20:21 soverysour left
[Coke] you can't call a shell function from raku's shell 20:21
patrickb $HOME/.rakubrew/bin/rakubrew internal_hooked Zsh ""
It's rakubrew complaining it's missing the shell hook. 20:22
[Coke] raku -e 'shell("/Users/coke/.rakubrew/bin/rakubrew internal_hooked Zsh")'
patrickb With the above command you call it just like the hook itself would do the call which should make the error disappear. 20:23
To make rakubrew work without the shell hook you need to switch to shim mode.
rakubrew mode shim
Then calling `$HOME/.rakubrew/bin/rakubrew` without the "internal_hooked" stuff should not error. 20:27
github.com/Raku/App-Rakubrew/blob/...rew.pm#L82 20:29
That's the place where that message is emitted. 20:30
lucs patrickb: Will try that out, soon (neighbour having problem I need to help with). 20:33
patrickb lucs: Having reread your original question, I think you have two options: using shell mode, or replicating the shell hook in your raku code.
But do note, that rakubrew is technically unable to tweak your PATH when you call it from raku code. (That's basically what the error message is about.) 20:36
When you want to keep env mode and just want to `download`, or reconfigure rakubrew or some such, then passing `internal_hooked Sh` should do the trick. 20:40
It's an interesting use case. 20:41
I'm afk again. Good night everyone!
[Coke] realizes he missed a bunch of earlier sends from lucs, so probably sounded very tone deaf in what he was saying - sorry about that. 20:57
lucs [Coke]: Hey, no problem. 21:05
Turns out that "you can't call a shell function from raku's shell" is pretty enlightening actually.
[Coke] \o/ 21:06
lucs And if I go back to the 'x' part of what could be called my 'xy' problem, the only functionality of rakubrew that I'm needing is to be able to shell out and run a specific version of Raku. 21:14
And that is easy to do by tweaking the PATH from with the Raku code, based on the RAKUBREW_HOME envvar.
21:17 orangebot left
timo I honestly would have expected you can call a shell function with the shell command. but if it's something like "exec" to update some env vars, it wouldn't do you much good since the shell it ran in will just exit right after that, undoing all your work 21:41
[Coke] well, looks like most of us are using zsh, and shell() doesn't use that, does it? 21:42
right, it's hardcoded to /bin/sh -c 21:43
21:46 johnjay joined
timo "you can't call a zsh function with rakudo's /bin/sh" doesn't really sound like something we can do much to fix :P 21:48
switching rakudo's "shell" over to use $SHELL would be a fantastic way to make everything break for every user of not-sh-compatible shells in any case 21:49
i remember when i changed my login shell to fish and couldn't start my desktop environment any more because something somewhere decided to try using my login shell to run some of its scripts during startup
david7832 Any one here have an idea on this (multi dispatch based on container type)? irclogs.raku.org/raku-beginner/2026-04-02.html 22:02
timo i think you might want a slurpy array parameter here 22:09
m: sub withslurp(*@a) { .raku.say for @a }; withslurp((1,2,3)); withslurp($[1,2,3]); withslurp([1,2,3])
camelia 1
2
3
$[1, 2, 3]
1
2
3
timo here the middle one has just one element when doing the for loop 22:10
david7832 Good idea. But I guess then every case will have to be handled by that subroutine? Because once I add a multi with signature ($a), this scalar multi almost always wins over the slurpy multi: 22:19
m: multi sub bc($a) { print 'have $ - '; $a.say; }; multi sub bc(*@a) { say 'have *@: '; .say for @a }; bc([1,2,3]); bc($[1,2,3]); bc("hello"); bc(1,2,3);
camelia have $ - [1 2 3]
have $ - [1 2 3]
have $ - hello
have *@:
1
2
3
david7832 Actually, I think you nailed it. This combination works very nicely: 22:28
sub esc-one($elem) { '"' ~ $elem ~ '"' }
sub esc(*@a) { do esc-one($_) for @a }
m: sub esc-one($elem) { '"' ~ $elem ~ '"' }; sub esc(*@a) { do esc-one($_) for @a }; say esc(5); say esc([1,2,3]); say esc($[1,2,3]); say esc([[1,2],[3,4]]); say esc($[[1,2],[3,4]]) 22:30
camelia ("5")
("1" "2" "3")
("1 2 3")
("1 2" "3 4")
("1 2 3 4")
22:44 Sgeo joined 23:21 wayland76 joined 23:32 hulk joined 23:33 kylese left 23:40 david7832 left 23:51 lichtkind__ left 23:53 runesicle left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!