🦋 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.
tbrowder i hope my solution works. created an empty file with the desired name. then i used that file name for the .save method 00:13
00:21 arkiuat left
tbrowder bummer, no go 00:29
01:31 annamalai left 01:57 sibl joined
simon_sibl so I need to run a program like cat to be able to do this ? 02:01
02:11 hulk joined 02:12 kylese left
[Coke] no 02:15
that example is passing your string *into* cat 02:17
so instead, pass it into the thing you want.
02:53 annamalai joined 02:56 annamalai left 02:58 annamalai joined 03:15 hulk left, kylese joined 03:51 sibl left
simon_sibl Oh yeah right sorry didn’t read well 04:05
I was just wondering since it’s possible to pass a opened file to :in and :out if it’s possible to pass a scalar instead of a file handler when calling run/shell directly 04:06
04:30 lichtkind__ joined 04:32 lichtkind_ left
Voldenet I've checked - :in(IO::Handle) just binds stdins, :in(True) just create IO::Pipe, :in("-") just uses stdin, :in(Anything else) does nothing 04:41
…am I reading it right? github.com/rakudo/rakudo/blob/c381...akumod#L47 04:42
05:43 kjp left 05:44 kjp joined 05:45 kjp left 05:46 kjp joined
simon_sibl idk but in the doc: docs.raku.org/routine/run 06:06
they have this line: my $proc = run "ls", "-alt", :out($ls-alt-handle);
but its an opened file,
06:18 wayland76 joined
Voldenet this has more info on how to use :in docs.raku.org/type/Proc 06:37
tbh. what you describe could be possible and not difficult to implement - the line I've posted would simply need to write the string to the stdin 06:38
07:05 sibl joined 07:50 annamalai left 08:01 Aedil left 08:05 abraxxa joined 08:07 Aedil joined 08:56 Sgeo left
simon_sibl yeah I guess I will just go for the print 09:07
I have a weird behavior with any module I try to install using Envy: termbin.com/xlcn 09:08
I am on Rakudoâ„¢ Star v2025.12.
I cant install any module
also I am not totally sure to understand the inst# and Envy# kind of path 09:09
09:13 dakkar joined
I see the last commit was 4y ago, is there an alternative ? 09:20
10:20 sibl left 10:28 abraxxa left 11:32 abraxxa joined 11:50 annamalai joined 11:59 annamalai left 12:00 annamalai joined 13:09 annamalai left
tbrowder if i use "run" to execute some command that creates an output file with just a base name, where will it exist on the local file system? 13:29
lizmat on whatever the OS thinks is the current directory 13:46
tbrowder ok, that's what i thought. i was confusing myself by keeping the output file name the same, duh 13:53
on 2 diff tests
13:59 abraxxa1 joined 14:01 abraxxa left
tbrowder when i use "run" in a test on github, how can i direct where the output file will go? i have tried specifying a temp directory, ensuring perms are 0o777", and i get a failure msg about being unable to open the file or directory 14:05
the same thing works in the temp dir on my local host 14:06
14:08 annamalai joined
tbrowder trying one more tweak... 14:17
ah, chatgpt to the rescue...trick with the :out... 14:24
[Coke] doesn't know what envy is. 14:25
[Coke] reads up 14:29
tbrowder well, getting close but no cigar... 14:36
antononcube @Coke I think "Envy" is programming language for doing Number theory. 14:47
Agh, well no, it is called "Sidef" and it is implemented in Perl... 14:50
[Coke] envy is on raku.land - looks like a zef wrapper to manage install locations. 14:52
by tony-o 14:53
tried to init a repo and install something to it, and it said it was already installed. 14:55
(init the repo, enable it, add the shim to zsh, zef install --to it... all candidates installed) 14:57
simon_sibl - it's telling you in the error message that you have to setup that repo in your LIB path for raku - with instructions for bash and zsh. 15:08
(did you then do it once in the current shell? ok) 15:09
Are you sure Event::Emitter works with just 'zef install <module>' ? 15:10
ugexe i don't think :in("some string") is really ever going to work well given "-" is already a magic string 15:12
Proc is also built on proc::async so i'm not sure even implementing a proof of concept would be easy or fit in the code well 15:13
i guess it could potentially turn the string from :in(Str) into the appropriate pipe object inside of BUILD or something 15:15
still wouldn't be able to pass in :in("-") though 15:17
some might find that acceptable though, similar to how IO::Handle.new(path => '-') opens STDIN and not a literal file named "-" 15:19
15:20 Sgeo joined 15:27 wayland76 left 15:28 wayland76 joined
[Coke] antononcube - I tried to comment on the youtube video about Math::Zeckendorf; I'd submit the comment, and then find it was gone a day later. did again today. submitted a link to raku.land - it showed up in my comment history immediately. Submitted one with a link to github - didn't even show up in history. now the raku.land one is also gone from my history. 15:32
a separate comment from yesterday wondering what was going on with my comments is still there, so it's not that I'm blocked from commenting on their video. :) 15:33
antononcube @Coke Where did you place those comments? At YouTube.
[Coke] raku.land/zef:coke/Math::Zeckendorf
er
www.youtube.com/watch?v=S5FTe5KP2Cw
antononcube Agh -- ok, at YouTube. YouTube has strange policies about posted links. 15:34
[Coke] interestingly, it still shows my comment with "1 reply" but now clicking to expand that hangs. :)
old top level ones vanish, replies are left in a zombie state. 15:35
antononcube Obviously, this means you should make your own video on the subject.
[Coke] were I already a youtuber, I would consider it.
... crap, technically I *have* uploaded youtube videos, but they are all private. 15:36
antononcube Private or unlisted?
[Coke] ... probably unlisted, actually
antononcube I will make the PR with document-templates to "Math::Zeckendorf" today or tomorrow. 15:37
[Coke] ok, sure
antononcube - Do you do any Literate Programming? - Pod6, Markdown, etc - Do you use Jupyter? 15:39
For Number theory LT can be easily done with "Text::CodeProcessing" if no graphics are involved. Using Pod6, Org-mode, or Markdown. 15:43
With graphics -- which I think makes anything Number theory more attractive -- the LT "weaving" (say, HTML, rendering) becomes more complicated. 15:44
[Coke] not literate. I do try to include docs in most non-toy things. 15:47
antononcube Ok, I will include workflow suggestions in the PR. 15:48
[Coke] ok
antononcube Needless to say, feel free to ignore and disagree. 🙂
[Coke] of course. :) 15:50
antononcube BTW, the most popular post -- in the last 3 years -- of the Raku Advent Calendar is Number theory related. 15:51
15:54 itaipu left
@Coke I forked it. Please, add tags and the raku.land link as shown here: 15:57
cdn.discordapp.com/attachments/633...15e13&
BTW, it is interesting that GitHub has tags like "zeckendorf" and "zeckendorf-representation", 16:00
16:06 abraxxa joined 16:08 abraxxa1 left 16:58 abraxxa left 17:00 derpydoo joined
[Coke] fixed the about 17:05
FYI, github's name for that is "topics", presumably since tags means something in git. 17:06
tbrowder leont: see my response to the Windows issue for App::Prove6 17:33
17:39 dakkar left 17:40 librasteve_ joined
librasteve_ lizmat: is this you? who.is/whois/raku.foundation 17:41
I just realised that rakufoundation.org forwards to the tprf site 17:42
leont tbrowder: could it be that the problem is that windows doesn't handle modules from CPAN, and it's resolved when I finally move Getopt::Long to fez? 17:43
17:44 snonux left 17:46 snonux joined
lizmat the former is me, the latter is TPRD afaik 17:48
*TPRF
librasteve ok - tx! 17:49
tbrowder could be, i'm not a Windows user and definitely not a win dev
[Coke] need me to test something on windows? 17:53
17:53 snonux left
tbrowder imho, it's not worth your time. the test provided by zef is probably all a win person would need unless we have a horde of non-mac non-linux ppl knocking at the door 😄 17:54
17:55 snonux joined
tbrowder though that would be nice 17:57
[Coke] app::prove6 install fine here, no errors about curl
if you don't have curl... you might have a very old or broken windows?
17:57 snonux left
[Coke] It's still unclear to me from the app prove6 ticket what exactly failed for you - that error output doesn't include the command that generated it 17:58
tbrowder the only windows i'm using is on github
antononcube @Coke See this Jupyter notebook: github.com/antononcube/Raku-Math-Z...tion.ipynb
Unfortunately, GitHub does not render JavaScript plots. I will make the Markdown version soon.
17:59 snonux joined
[Coke] tbrowder: can you point at the action that's failing? 17:59
what version of windows is it?
tbrowder i'll see... 18:00
librasteve weekly: perlfoundation.org/fosdem/community-dinner.html 18:01
notable6 librasteve, Noted! (weekly)
tbrowder my prove6 line may be wrong. i couldn't find a good example of the simplest use 18:02
librasteve ^^ anyone connected to perl||reaku is welcome to join the FOSDEM community dinner
antononcube Is the dinner online or in-person?
[Coke] Did you miss a /s tag? 18:05
In general, and I say this a lot: it's hard to diagnose a problem if we can't see what's failing
well, not those words exactly, but the sentiment persists. :)
tbrowder i know i know 18:09
see i don't even know what an /s tag is :( 18:10
[Coke] that was for antononcube - sarcasm. 18:11
tbrowder are we talking about workflows?
antononcube @tbrowder_04810 I asking is FOSDEM community dinner online. 18:13
tbrowder with an yml file in workflows, what should the "run prove6" line look like?
18:13 snonux left
tbrowder according to its docs i see "run prove6" 18:14
librasteve surely sarcasm needs to be funny? 18:15
tbrowder no other input. it's supposed to find t (and other std test directories) and run those tests. afk& 18:16
antononcube @librasteve I think some sort of Poe's law manifestation is in place... 18:19
[Coke] sarcasm being funny is a nice to have, unfortunately 18:24
18:32 derpydoo left
librasteve ;-) 18:35
18:37 snonux joined
rcmlz I am tryint to install Rakubrew on my Mac M4 and get: Couldn't find a precomp release for OS: "linux", architecture: "arm64", version: "2025.12" 1.483 You can try building yourself. Use the `rakubrew build` command to do so. Is that expected, that there is not build for ARM64? 18:58
[Coke] I believe that worked fine on my M2.. 18:59
did you use 'rakubrew download moar 2025.12' ? 19:01
... wait, linux?
rcmlz I did now rakubrew build moar-2025.12 and that worked finally
It is in Docker, FROM jupyter/datascience-notebook
[Coke] Are you running macos or linux? 19:02
rcmlz macos
[Coke] ok - there is a macos version. Why does your error message say "linux" ?
are you trying to build it *inside a linux docker container* but on a mac host? 19:03
rcmlz Yes 19:04
is that wrong? I am just started docker journey
[Coke] .. ok, that's going to look for a _linux_ version, not a mac version. and no, I don't think we have linux arm64 builds pre-done, only macos arm64 builds. 19:05
if you want a mac version, run rakubrew from the host OS.
So, not wrong, per se 19:06
rcmlz Good to know. The rakubrew build moar-2025.12 worked, so all fine - was just surprised that that one is missing
[Coke] I think having to build is the correct answer there - or use a pre-existing docker container that already has raku installed, but I don't think anyone is necessarily generating a new one each compiler release 19:07
I believe only one flavor of linux is pre-built.
flavor==hw arch.
glad you got there! 19:08
tbrowder [Coke]: do you really want the whole github output from the prove6 failed test on github windows? or is just the yml file good enough? 19:09
it’s a LOT of nothing burger until the fail-to-find line when it aborts with the msg i showed 19:12
[Coke] can you just link the report? 19:16
yml file is a good start.
I'm also wondering where that native curl item not found is coming from. 19:25
... it's not in App::Prove6
[Coke] wonders how long App::Zef-Deps has been broken. 19:28
compilation error on mac, VERY long hang on windows. (but eventual success) 19:29
yah, nothing in App::Prove6's hierarchy depends on native curl. So I assume that's one of your own deps on whatever it is you're testing 19:30
leont: sigpipe tests fine on windows. 19:31
japhb just did a coding interview in Raku 19:50
I love it when they let you choose your language to do it in
[Coke] Nice, good luck! 19:57
"is psuedocode ok?" "great, <uses raku>"
tbrowder: I have a minute between calls here if you have the yml file... 19:58
antononcube @japhb For what kind of job(s)? 20:02
[Coke] ugexe: with zef install --debug App::Zef-Deps on my box, seeing these two lines: 20:03
[App::Zef-Deps] ===SORRY!=== Error while compiling /private/var/folders/8n/rypl9_990b9578zd5435pgr40000gn/T/.zef.1768507379.18985/c92e0e2dd6275
[App::Zef-Deps] 905077113f9ce9ede04cf50378c.tar.gz/App-Zef-Deps-0.9.8/t/basic.t
japhb antononcube: Staff SRE 20:05
[Coke]: Yep, that's pretty much it. In fact the interviewer asked why I chose Raku, and I told him when I write pseudocode, it's pretty close to executable Raku anyway, so .... 20:07
[Coke] japhb: heh 20:08
antononcube I have to say in Data Science (DS) and LLM/AI fields, Python is imposed. If I even mention that Python is not that good DS or ask "Why Python?" this makes managers, recruiters immediately but "black marks." 20:10
I am not is that good news or the bad news, BTW.
[Coke] has pondered adding a "learning python" class for CPE credit this year, should probably
antononcube The problem is that Python is somewhat good for DS, and its libraries are fairly "illogical" at times. Because of this there is a large body of texts that tell how to do certain DS things. Hence, LLMs "know it." 20:12
For example, it is not hard to produce training exercises that can be used to prepare for certain interviews. 20:13
Hmm... this might be a good Raku blog post: «Making code interview exercises». (For any language and domain.) 20:17
japhb antononcube: :+1: 20:23
20:23 japhb left
tbrowder [Coke]: coming at cha... 20:26
when i find it again...found it, pushing again to get a check before i gist it 20:41
20:42 japhb joined
tbrowder www.irccloud.com/pastebin/goUgNNZF/ 20:53
same error.... 20:57
i still like using zef for workflows. prove6 is very heavy and i tend to write special tests for one-off features. they can go in xt if need be, but i like to see them in action on the other two OSs. 21:01
oh, and mi6 takes care of running all tests 21:05
21:25 sftp left, sftp joined, sftp left, sftp joined 21:29 librasteve_ left
ugexe i dunno what your issue is, but `--exclude="z, curl"` is not how to exclude two items. that would be `--exclude=z --exclude=curl` 21:53
21:53 wayland76 left
ugexe --exclude="z, curl" would ignore a module called literally "z, curl" 21:57
m: class ::("z, curl") { }; say ::("z, curl"); 21:58
camelia (z, curl)
ugexe m: class ::("z, curl") { }; say ::("z, curl").new;
camelia z, curl.new
ugexe in raku you pass multiple values (i.e. arrays) by using the parameter multiple times, e.g. --exclude=z --exclude=curl 21:59
22:08 atcroft left
[Coke] tbrowder: can you not send us a URL -to the actual yml file in situ-? 22:38
so then I could see, for example, what tests you're trying to run?
and I could fork the repo, test changes, etc?
and you mention earlier, you like using zef.. why not run zef without --/test and use zef to test it? 22:39
but definitely fix ugexe's comment, then maybe share the link, then consider the other items. 22:40
but without seeing the code you're testing, I can't tell you why it's complaining about curl.
ugexe the one reason i can think of is to run author tests 22:41
github.com/ugexe/zef/blob/2c6ab9ff...ml#L26-L30 for zef i just run the xt tests manually in the ci pipeline
22:45 Aedil left
[Coke] looks like github.com/tbrowder/PB-Lottery/blo...prove6.yml 22:54
tbrowder yes 22:56
seems to me that what it's testing is unimportant, it's trying to get App::Prove6 loaded. that same module tests fine on windows fine using zef 23:05
instead of prove6
the module is my PB::Lottery 23:07
branch next-ver 23:08
ugexe "Download of rakudo-star-2022.07-01-win-x86_64-msvc.msi (24.96 MB) completed."
i have no idea why you are installing rakudo star at all 23:09
but that isn't going to help
it isn't even making it to the prove6 command 23:10
===> Failed to find dependencies: curl:ver<4>:from<native> 23:11
probably installing rakudo star from 2022 is installing a old as hell prove6 with all sorts of issues
if you have indeed made it that far previously 23:12
[Coke] oh, ew. 23:13
23:13 atcroft joined
[Coke] Good catch. 23:13
tbrowder i don't usually try to update windows things that still work. i am trying to give potential users whether it works on windows or not 23:15
[Coke] Note that curl *is* available on whatever image you're using: curl 8.16.0 (Windows) libcurl/8.16.0 Schannel zlib/1.3.1 WinIDN
ugexe yeah but the binary detection for zef isn't the greatest so you often have to --exclude various native library/binary things if you know you indeed have them 23:16
you dont need to install rakudo star. you are using a github action image that already has rakudo installed 23:17
tbrowder once again, i don't have windows. i sometimes have to fiddle with the yml file to get it to pass on github. once it does i don't touch it. 23:20
zef "just woks" for me 23:21
[Coke] this isn't a windows issue.
tbrowder *works
[Coke] I mean, you're running it on windows, yes, but there's extra stuff in here that doesn't need to be here. I have one that is installing all the dependencies right now. one moment. 23:22
tbrowder but what difference does that make?
[Coke] what difference does "what" make? 23:23
tbrowder having extra stuff
[Coke] ok, I have a version that is now complaining that your tests don't pass. 23:24
... it slows it down, it could put things in your path that *don't work*...
Question: Do you really want the tests run via App::Prove6, or was that just to try to get this working?
tbrowder great, no one else uses 3-OS testing so how would i knwn if the owner is testing on windows or not without looking at his test.yml file? 23:27
ugexe i use 3 os testing
github.com/ugexe/zef/blob/2c6ab9ff...ts.yml#L10
tbrowder yes but yr zef is in the stratosphere 23:28
if zef doesn't work we all go back to the stone ages :) 23:29
[Coke] Question: Do you really want the tests run via App::Prove6, or was that just to try to get this working?
aruniecrisps does anyone know of any raku packages where the source is hosted on codeberg? 23:31
[Coke] Here's a run in my fork that passes your tests: github.com/coke/PB-Lottery/actions...0533706603
tbrowder that was to show why i don't use prove6 and to give @leont some info on my issue with App::Prove6
aruniecrisps i'm thinking of either migrating my current repos there/creating my repos there from now on
[Coke] If we skip App::Prove6, can probably trim down the runtime there to faster than 2.5m
there is no issue with App::Prove6 - it's working. 23:32
ugexe aruniecrsps: grep 360.zef.pm/index.json for codeberg
[Coke] So you can use it on windows if you like. I'll clean up my branch here and submit a PR you can take if you like.
aruniecrisps awesome! thanks! 23:33
23:33 elcaro left
thanks! ugexe 23:35
tbrowder ok. but i wish we had a place to have os.yml files available. at least ones that you guys approve of. mixing all os stuff in one file to me is not useful u
*unless* they are guaranteed to work on all 3 os 23:36
[Coke] tbrowder: github.com/tbrowder/PB-Lottery/pull/2 23:37
I don't see why you need a different set of runs for your linux and windows versions. 23:38
the main difference at this point is the use of prove6
but you could use that in both if you wanted. (or neither)
and your linux & macos differ only by the os name. 23:39
tbrowder because i want users to see if their OS is supported with the github badge
[Coke] You could easily have the same zef install --deps-only line for all three. 23:40
... note I didn't say "different action files"
You could have the guts of each file be identical, though.
ugexe they do it to have different badges for different OS. but i dont really agree that it solves a problem. first, if your code should work on all 3 OS then you shouldnt be commiting the broken code to the main branch if it isn't passing all three OS. For the case where the author *knows* their code doesn't work on a different OS but they still want to release it so other OS users can use it, running CI
tests on the code for the broken OS is a waste of resources and time 23:41
[Coke] Anyway, hope the PR helps.
ugexe and in that case the user can explicitly say it doesn't work on windows, they don't need a badge for that
i do concede you need separate files to have multiple badges. i just personally don't think in practice it solves the general problem of knowing if something is intended to work on windows or not 23:47

Powered by Cro and the Raku® Programming Language.

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

Thank you!