🦋 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.
00:00 reportable6 left 00:01 reportable6 joined 00:03 xinming left 00:07 jpn left 00:08 xinming joined 00:14 a3r0 left, a3r0 joined, DarthGandalf left 00:15 DarthGandalf joined 00:23 epony left 00:27 freeside left 00:32 Manifest0 left 00:40 freeside joined 00:44 cfa left 00:46 freeside left 00:49 freeside joined, jpn joined, epony joined 00:51 ProperNoun left 00:55 freeside left 00:59 rf_ left, rf joined
rf . 01:02
Evening folks, I won't be too active here and in the doc repo for a bit. Having some massive prod issues at work, and of course I'm on call. If you need me though i'll be checking IRC for tells everyday. 01:03
01:03 xinming left 01:05 xinming joined 01:12 freeside joined 01:17 freeside left 01:22 rypervenche joined 01:23 freeside joined 01:50 ProperNoun joined 01:53 MasterDuke joined
guifa rf, happens to the best of us sometimes. but appreciate all your work 02:01
02:12 jpn left
rf guifa: thanks! 02:13
02:36 jpn joined 02:43 jpn left 02:45 jpn joined 02:50 jpn left 03:01 jpn joined 03:08 jpn left 03:19 jpn joined 03:29 jpn left 03:53 freeside left 04:21 jpn joined 04:22 freeside joined 04:29 jpn left 05:14 jpn joined 05:19 jpn left 06:00 reportable6 left 06:03 reportable6 joined 06:08 jpn joined 06:10 freeside left 06:12 jpn left 06:47 freeside joined 07:03 jpn joined 07:08 jpn left 07:45 Sgeo left 08:28 jpn joined 08:33 jpn left, stevied_test left 08:34 stevied_test joined 09:04 Manifest0 joined 09:08 ab5tract joined 09:10 DarthGandalf left, dakkar joined 09:14 DarthGandalf joined 09:24 freeside left 09:28 freeside joined 09:30 sena_kun joined 09:32 freeside left 09:36 freeside joined 09:41 freeside left 09:58 abraxxa-home joined 10:01 abraxxa-home left 10:21 grondilu joined
grondilu Hi all. 10:21
So, some time ago I broke my Digest module. 10:22
Or worse : I made changes that broke modules that were depending on it.
I think it's kind of unfortunate that this can happen at all. 10:23
el reinstall it?
grondilu Lately I've learned about Nix/NixOS, and if I understand correctly with that system of package management it's impossible to break something. 10:24
Maybe zef could take inspiration from it to avoid that kinds of pitfalls?
10:28 freeside joined 10:32 freeside left
grondilu or even look into Hydra, NixOS's Continuous Integration system. Seems pretty cool. 10:33
in any case package management seems to me like quite a generic task, I'm kind of surprised that Raku needs its own system for it, can't it use an already existing, possibly more robust system? 10:34
lizmat grondilu: you broke it because you changed the code and not updated the version ? 10:40
grondilu I did change the version, I think. 10:41
lizmat if that's the case, I think it's at least partially the fault of the user for not pinning the version that they want to use in their production code 10:42
grondilu but the truth is, I find it quite annoying to have to change the version number all the time.
El_Che I would love to one day evolve to "build" of a minimal rakudo + all libs. Of course, C dependencies shatter the dream a little :)
lizmat grondilu: how would people be able to tell which version they want of your module?
from a sha? 10:43
grondilu yes
a digest
like in nix
El_Che grondilu: versioning is essential, and not moving tags
grondilu El_Che: I hate it
El_Che otherwise users will hate you
semantic versioning is where is at today
and that create expectations 10:44
s
Nemokosch well, "pinning versions" is only as much a solution as bundling someone else's code with yours, offline
lizmat grondilu: you could use something like App::Mi6 that basically makes updating version numbers something like "version 0.0.4? Y/n"
El_Che Nemokosch: bundling is good 10:45
tellable6 El_Che, I'll pass your message to Nemokosch
Nemokosch unless it turns out there was, say, a zero-day in what you bundled
receiving updates to something in production - even "implicit", mandatory updates - can be important 10:46
El_Che Nemokosch: bundling == you are responsable, you need to publish an update
tellable6 El_Che, I'll pass your message to Nemokosch
Nemokosch then it's basically not actual modularity
El_Che and it's not because it's bundled that you can not update it, or get an update lib before on the search path 10:47
Nemokosch: yes by design
tellable6 El_Che, I'll pass your message to Nemokosch
Nemokosch then probably the generalization that "this needs to be done in production" seems unfortunate
then probably the generalization that "this needs to be done in production" seems unfortunate 10:48
El_Che that's why ruby's ecosystem is a burning garbage container
everything breaks all the time
in gem land
Nemokosch anyway grondilu, there is a reason concepts like semantic versioning exist 10:49
grondilu I think it's ridiculous that a package management system can fail so bad that it breaks things.
Nemokosch not to say that I'm super good at authoring but I lowkey wish the ecosystem enforced some meaningful versioning, rather than Whatever
and a lot of the worms seem to be coming from the banal fact that a lot of modules were authored in a time when people were happy to publish Perl 6 code at all 10:51
versioning was nonexistent in the ecosystems, and that was the least thing to worry about
another problem that grondilu's example highlighted is the lack of conceptual awareness of the dependency resolution system 10:53
I was also shocked to learn how it is designed to work - and on some level, it doesn't matter if somebody is "shocked" or just "surprised" 10:54
grondilu I admit I don't know the details of how these things work.
Nemokosch if you think that your dependencies are dists and at some point it turns out that they are in fact modules, probably there has been enough harm done in the meantime
grondilu but I can see that how things work currently seems very far from optimal 10:55
10:56 abraxxa joined
Nemokosch grondilu: for the record this was a big part of the module breakage. if you depend on "Digest", that's Digest the module, not Digest the dist - however, the dist may also be found and resolved to it, as a kind of fallback 10:56
grondilu like the name Conflict there was with the Digest::HMAC module. This was ridiculous IMHO 10:57
Nemokosch there were two bites
in quick succession
the first bite was that dependencies are all modules, irrespective of the dist they are contained in - therefore your Digest::HMAC module can collide with any other Digest::HMAC module (or dist, at installation) 10:58
grondilu modules or repo should not be identified with a low entropy name such as 'Digest', IMHO 10:59
it's a recepe for name resolution conflicts.
Nemokosch I mean, why did you give it that name? 😄
it clearly increased the pressure
grondilu I mean the names should be salted 11:00
like they do in Nix. I'm impressed by this system, I think they do things right.
Nemokosch cont "modules are the dependencies": inb4 yes, if you pin the author, you half solve the problem - collisions are still possible within the same author but then one will blame the author anyway, right?
11:00 abraxxa left
el salt it yourself Digest_718719794791 11:01
Nemokosch 🤣
grondilu fair point
Nemokosch by the way, how does a dependency look like, in that system?
grondilu I think they just use hashes everywhere, basically. Not sure that answers your question, though. 11:03
Nemokosch yeah nah, doesn't give much of an idea about the user's perspective
anyway, let me finish the analysis - the second bite was that in the META6 file (and clearly not in the code!), a name still may resolve to a distribution rather than a module, and apparently module resolution doesn't even have absolute priority over dist resolution 11:05
this means that somebody could depend on "Digest", understood as a dist, and then try to use Digest - which wasn't provided by the dist anymore
grondilu but what does the 'depends' entry mean in the META6 file? Does it refer to modules or repos? 11:09
Nemokosch it means modules as a main rule - but zef at least can find dists as well, under some circumstances
grondilu I thought it meant repos, so I thought not having a Digest module in my Digest repo was fine. 11:10
Nemokosch well, if it meant modules strictly and consistently, that would have probably worked...
because then it wouldn't have treated a recent "Digest" as a possible candidate at all 11:11
also... in case anybody is already getting salty somewhere, about the "complaining". YES, the author of the module has made a mistake (improper versioning - changing modules of a distribution is a breaking change on the highest level, not sure if it should even be allowed tbh). 11:14
YES, the users also made a mistake (by not pinning their dependencies by anything else but name - not any pattern of version, not the author) 11:15
grondilu That's the thing : the system should be more robust and not break because of a human error.
At the very least, it should notice the upgrade failed and then rollback, no? 11:16
Nemokosch Well, if it treats modules the dependencies, it definitely shouldn't select the dist when there are modules with that name available 11:17
el if there is a name collision should that be resolved with namespaces? 11:19
or is it a different problem 11:20
Nemokosch "Always complaining", right? If by complaining, we really only mean criticism without offering possible directions of resolving the problem - then all the complaining is about the attitude of "fix yourself", that's basically a betrayal of the Perl 6 vision
For the dependency management, there have been several, really varied ideas for improvements, a great variety honestly 11:21
I don't know much about Nix and dependency immutability but that's also a point
"clarifying the relation of a distribution and a module" is like five points 11:22
this is about dependency resolution, my feline friend 11:23
grondilu I enjoy writing raku code, but having to housekeep a module repo is not fun at all to be honest. 11:24
el elaborate im lost in this discussion 😭
Nemokosch grondilu: I think it gets easier once you have the right idea of what is going on
however, I have to say I don't like "what is going on", and that I've already said with arguments that basically just bounced back without an acho 11:25
*echo
I still think not everything is lost and dependency management could be designed around distributions, rather than modules floating in heavens 11:26
lizmat grondilu: again, things like App::Mi6 do most of that work for you
grondilu lizmat: I'll look into it 11:27
lizmat and speaking as the author with the most modules in the ecosystem, I think I can say I know what I'm talking about :-)
Nemokosch however, if module dependencies are really the way to go (which puts immediate pressure on the naming of your exposed modules, too!) - then the META6 needs to get "nerfed" in favor of paying attention to metadata directly provided by modules 11:28
grondilu lizmat: I don't doubt it, but not everyone can spend a lot of time into figuring out how all of this work
Nemokosch hence basically making modules as sort of mini-distributions 11:29
I don't think this is a hard rule but right now, ecosystems rely a lot on META6 solely 11:30
and META6 exists per distribution, not per module
lizmat grondilu: in short: zef install App::Mi6; mi6 new Foo; cd Foo; edit lib/Foo.rakumod; edit t/01-basic.rakutest; mi6 build; git diff for checks; mi6 release
Nemokosch which practically means that even a properly-pinned name is "dirty": it partially refers to data tied to the distribution, and data tied to the module 11:31
el i think its about breaking something and forgetting to specify a version
grondilu maybe a bit out there but : shouldn't module distribution informations belong in POD6 comments inside the module, instead of a separate META6 file?
Nemokosch basically yes, I think that would be one way to get more stable and transparent 11:32
grondilu also then that would settle the issue about module vs repo distributions : everything would be modules, and a repo would just be a set of modules. 11:33
Nemokosch yes!
grondilu glad to see we agree
Nemokosch right now, META6 (and the lack of consideration for module-provided metadata) is the biggest enemy of that
mind you, modules can technically provide their metadata - it's just not taken into account over META6 11:34
grondilu if I'm not mistaken, at the moment POD6 is very much NYI, isn't it?
that is the %*POD6 var or something
Nemokosch good question but I'd say that would be an unrelated discussion. Modules can communicate their metadata with attributes 11:35
11:35 lizmat left
Those attributes are just ignored/not set, most of the time 11:35
grondilu looks the doc about pod variables 11:36
the = twigil is what I was thinking about
I guess that would be needed to put the distrib info in POD6 comments 11:37
Nemokosch another question would be: does it even make sense to say that a distribution has a "version"? And if it does, that version has loose relation with the module versions
grondilu 🤷 11:38
Nemokosch let's take your Digest distribution again. Say, you make a change to, idk, your Digest::MD5 module. Then clearly it receives a new version. However, if you have, for example, a Digest::MAC module as well, why should that get a new version? 11:39
moritz usually just for simplicity 11:40
grondilu well it shouldn't if we decide to distribute everything as modules.
Nemokosch I mean I don't think publishing the same thing under different consecutive versions can actually break anything but it's confusing at best
Yes, well... another banal solution is to just not bundle any two modules together 😆
Then a module will indeed correspond to a dist 11:41
moritz if you have shared code between your modules (in the form of helper modules, for example) it can be really hard to tease out which module is affected
and if you don't have any shared code, yes, separate distributions sound like a good idea
Nemokosch I was about to say that moritz probably knows the history of distributions better 11:42
Apparently it's pretty much the same thing in Perl 5 land
11:43 lizmat joined
perhaps it seemed safer to do "whatever changes, everything changes" 11:43
because of the possibility of internal changes, not exposed by the module 11:44
11:54 grondilu left 12:00 reportable6 left 12:02 reportable6 joined
anyways... actually I had a question - does anybody have experience with generating module code as a part of the build? 12:03
Would make life easier for HTML tags
lizmat whenever I needed to generate code, I usually wanted to have that as a separate step for the developer, using a helper script 12:07
if you really want that, I guess using a Build module would be the place ?
Nemokosch yes, I would think so, it's just I never did something like that so I don't really know the places of interaction, like what can happen when 12:12
this Build module would run before the usual installation step? 12:13
12:15 woei joined
ugexe what language are you using where every module in a bundle is versioned searately? 12:47
Nemokosch this is kind of a "principle of explosion" question without first asking: what language makes modules of a bundle with no individual metadata, the units of dependency 12:49
ugexe ah, the 'ol answer the question with another question 12:50
Nemokosch it was begging the question 12:52
ugexe is that what begging the question means? 12:55
jast begging the question essentially means circular reasoning 12:56
ugexe right 12:57
jast more specifically, assuming the conclusion
12:58 ispyhumanfly left
jast example from Wikipedia: "Green is the best color because it is the greenest of all colors" 12:58
Nemokosch sorry if I didn't use the right term, maybe "complex question" fits better en.wikipedia.org/wiki/Complex_question 13:01
ugexe alright, then: ah, the 'ol answer the question with a complex question 13:04
lizmat ugexe: why do you say that? :-) (no, don't answer :-)
Nemokosch more like, the ol' "ask a complex question out of context to imply something" 13:05
lizmat perhaps we should get back to the question: does anybody have experience with generating module code as a part of the build? 13:06
ugexe ok. well my point is (since i've recently mentioned "answering questions with more questions" here) that answering questions with more questions (regardless of if they imply something, or whatever) isn't likely to get you answers
El_Che isn'it? (back to the circular thing :) )
ugexe there are a couple of modules that had like Foo.pm6.in file (a template) and filled them in at build time
i don't remember them though 13:07
Nemokosch anyway, this is a usual case of, people will react to what they want, the way they feel like. Apparently the discussion was more worth adding questions of dubious value than sharing experience about codegenned modules 😦 13:08
ugexe no, its just that you have a pattern of answering my questions (and probably others) the same way 13:09
i'm trying to help you get your answers, but i'm not a chatgpt bot 13:10
Nemokosch and you have a pattern of asking presumptuous questions, and also declaring at one point that the whole discussion is just pointless. Why even bother, then, really? Who does it help? 13:11
ugexe well, have it your way i guess. I won't keep wasting both our time 13:12
Nemokosch tbh IRC probably isn't even the best place for certain discussions to happen 13:17
however, unfortunately there is this tendency that on github, one must have tons of luck to get anyone's attention for a topic 13:18
p6steve I would like to have support for code generation in the module build-execute process ... the use case is (i) there's a raku slang, (ii) it can parse and codegen for a compiled language (C, C++, Rust...), (iii) it can dynamically build a SO and (iv) it can then call the new SO to execute the new code --- where is the best place to post this as a feature requiest? 13:26
ugexe you can do all of that 13:29
Inline::Perl5 does this
you essentially just do all of that in a make file and then execute that in a Build.rakumod or using Distribution::Builder::MakeFromJSON 13:32
p6steve sorry --- I did not mean to imply that this can't be done --- thanks for the pointer
Geth doc/lizmat-lose-compose: 47a9da7df3 | (Elizabeth Mattijsen)++ | doc/Language/phasers.pod6
Lose mention of COMPOSE phaser

Associated with github.com/rakudo/rakudo/pull/5193
Note: there are *NO* tests for the COMPOSE phaser		 13:37
13:52 jpn joined
ab5tract Nemokosch: Can you clarify what you mean about "making life easier for HTML tags" ? 14:04
tellable6 ab5tract, I'll pass your message to Nemokosch
ab5tract I wrote a recent blog post about generating subs on import, not sure if that is something that could help 14:05
Nemokosch I'm thinking about getting decent coverage of tags in the HTML::Tag dist 14:15
(because, well, I noticed that the poor guy continuing my work is doing ugly hacks and it's simply because HTML5 tags are usually missing, lol) 14:16
so one thought I had is that I don't mind having dummy classes for each and every tag; that type system can come handy possibly 14:17
Geth ¦ doc: coke assigned to MonalikaPatnaik Issue update CONTRIBUTING.md github.com/Raku/doc/issues/4189
Nemokosch however, in that case, it would be good to generate these dummy classes, rather than having, I don't know, 100 in a maintained raku module file
ab5tract Hmmm, defining classes on the fly like that is probably a lot trickier than subs 14:19
Though in either case my approach would probably avoid actual codegen in favor of dynamically generating definitions in code 14:20
I guess that's the kind of thing that RakuAST is going to unlock ...
rf Lol prod issue at work caused almost 2tb of memory to leak in 5 hours 14:27
We figured it out though :D 14:28
El_Che rf: you should close DB transactions! 14:29
:)
rf It was actually S3 bucket files lol
The developer thought the file handles were being closed but they definetley weren't
El_Che close 14:30
Nemokosch ab5tract: you know, I'm fine with simply generating the file, as source
it's a very simple structure, after all 14:31
14:34 perlbot left, simcop2387_ joined, simcop2387 left
ab5tract Not sure how to do that in a way that "just works" 14:34
14:35 freeside joined, simcop2387_ is now known as simcop2387 14:38 perlbot joined
moritz cd 14:38
sorry
Nemokosch 😅 14:39
either way, it's good to see you around every now and then 14:40
moritz even when I'm just here for changing my directory :-) 14:42
Nemokosch a rare feature of the #raku channel 14:45
rf lol
Nemokosch pfff, thank you, discord
# r a k u
rf What does :D: vs just :D mean? 14:47
In type definitions.
Nemokosch : appears in methods in particular, no? 14:48
in which case I'd assume you mean the separator of the "self" parameter
rf Something like method(Complex:D: --> Complex:D)
Nemokosch yes, that's for the "self" 14:49
rf I see
Nemokosch it can be called on a Complex:D and returns a Complex:D
fun fact: there even is a call syntax like that
instead of foo.bar(baz) or foo.bar: baz, there is even bar foo: baz
moritz TBH after doing Python for quite some time, I like their default better, which is that all methods are instance methods, unless you specify it otherwise 14:50
rf submethod vs method? 14:51
moritz forgetting the :D on the invocant is just too easy, and when you specify it all the time, it's quite the clutter
Nemokosch method vs "class method"
moritz submethod is not inherited, that's orthogonal to class vs. instance method
Nemokosch It's not even that clear what would be THE one "class method" in Raku
oh well, thinking Python, it is clear after all 14:52
a subroutine scoped in a class - that's a "static method"
a method taking a type object - that's a "class method"
perhaps?
tbh I don't know many languages that make this distinction, let alone as consciously as Python does 14:53
moritz most "static" OO languages make that distinction 14:58
"static" in java and C++ (iirc)
(where with the first "static" I meant statically typed)
Nemokosch as much as I remember, C++ only had "static methods" by Python's terms 14:59
which would make sense since C++ classes are not first-class citizens of the language
so a function cannot just take them
I'd assume the same about Java, C# etc even though they have reflection some way 15:00
15:20 jgaz joined 15:40 zashi joined
Voldenet it would be awesome if there was no distinction between static and not on syntax level 15:46
where using reference to self would promote the static method into instance method automatically 15:47
not that it's realistic to implement that across module usages
15:48 Sgeo joined
Nemokosch well, this is Raku, if you don't mind the risks 15:48
that you really meant a definite instance and ended up passing a type object
Voldenet m: class { my $.ok = 1; method instance-method { say self.ok; } }.instance-method 15:50
camelia 1
Voldenet Yes, this is proper code
forgetting the .new gives you runtime error so it's not that amazing 15:51
rf why my over has here? 15:52
Voldenet :> 15:53
my is a static variable here
m: class { has $.ok = 1; method instance-method { say self.ok; } }.instance-method; 15:54
camelia Cannot look up attributes in a <anon|1> type object. Did you forget a '.new'?
in method ok at <tmp> line 1
in method instance-method at <tmp> line 1
in block <unit> at <tmp> line 1
Voldenet tbh though, it's poor syntax in this case, really implies something else 15:55
m: class { my $ok = 1; method instance-method { say $ok; } }.instance-method;
camelia 1
Voldenet this is more obviously static
Nemokosch m: my $.ok = 1; dd $.ok;
Raku eval Exit code: 1 ===SORRY!=== Variable $.ok used where no 'self' is available at /home/glot/main.raku:1 ------> my $.ok = 1; dd $.ok⏏; expecting any of: argument list term Other potential difficulties: Useless generation of accessor method in mainline at /home/glot/main.raku:1 ------> my ⏏$.ok = 1; dd $.ok;
Nemokosch gotcha
still... is that $ok only an accessor or a normal, fully-fledged variable? 15:56
m: class { my $ok = 1; method instance-method { say self.ok; } }.instance-method;
Raku eval Exit code: 1 No such method 'ok' for invocant of type '<anon|1>' in method instance-method at main.raku line 1 in block <unit> at main.raku line 1
Nemokosch it is a variable, then
with has scoping, it would be just an alias of the accessor 15:57
my $.ok could even be an oversight?
Voldenet still, using accessors on type objects feels weird
Could be, but maybe generics change something here 15:58
Nemokosch Again, I'd hope that the new compiler frontend will be considerably cleaner 16:00
hence easier to reason about these decisions
see why they came to be and how they could be changed
Voldenet nope, generics don't change anything in here 16:01
m: role n[\T] { my $.ok = 1; method instance-method { self.ok++; } }; for ^10 { say n[$_ %%2 ?? Int !! Str].instance-method } 16:03
camelia 1
1
2
2
3
3
4
4
5
5
Voldenet m: role n[\T] { my $ok = 1; method instance-method { $ok++; } }; for ^10 { say n[$_ %%2 ?? Int !! Str].instance-method }
camelia 1
1
2
2
3
3
4
4
5
5
16:14 freeside left 16:22 freeside joined
Scotteh lizmat: Thank you for the lovely message on the obituary. 16:24
16:34 camelia left, m_athias left, nine left 16:35 nine joined, m_athias joined 16:38 m_athias left, nine left, m_athias_ joined 16:39 camelia joined 16:40 nine joined, zashi left 16:43 m_athias_ is now known as m_athias 16:50 jpn left 17:30 Xliff joined 17:38 dakkar left 17:43 freeside left 17:44 freeside joined 17:53 freeside left 17:54 freeside joined 18:00 reportable6 left 18:01 reportable6 joined
rf Been thinking about writing an emacs package for fez 18:01
18:05 freeside left 18:09 nine left, m_athias left 18:10 nine joined 18:19 freeside joined 18:25 Monalika9 joined 18:29 jgaz left
lizmat Scotteh: you're very much welcome 18:32
19:10 epony left 19:11 epony joined 19:20 freeside left 19:23 freeside joined 19:27 jgaz joined 19:28 freeside left 20:01 NemokoschKiwi joined, freeside joined 20:23 xinming left 20:25 xinming joined 20:37 Monalika9 left 21:01 freeside left 21:05 NemokoschKiwi left 21:12 freeside joined 21:16 deoac joined, freeside left
deoac Is it possible to use Perl source filtering in a Raku program?  I've tried ```use Filter::Simple:from<Perl5>;``` without success. 21:17
21:25 melezhik joined
melezhik .tell jjatria I created a simple CI pipeline that also runs a code snippet from http tiny readme, let me know if you are interested and I can create a pr 21:26
tellable6 melezhik, I'll pass your message to jjatria
p6steve deoac: let me test on my kit 21:30
what is your OS? how did you install rakudo & Inline::Perl5? 21:32
21:34 melezhik left 21:48 freeside joined
deoac: it works for me (although I barely know how to use this module 21:49
gist.github.com/p6steve/624fe58c68...7d35068496
^^^ here's how - you are welcome to use or to plunder my example Dockerfile 21:50
===SORRY!=== Error while compiling //scum.raku Unsupported use of /g. In Raku please use: :g. at //scum.raku:4 21:51
21:51 jpn joined 21:53 freeside left 21:57 jpn left 22:00 jpn joined 22:06 jpn left 22:07 sena_kun left 22:08 freeside joined 22:14 deoac left 22:18 Xliff left
rf Can I destructure a Pair the same way we can a List 22:32
Something like
m: my $pair = :abc(123); -> :$key($value) { say $key, $value; }($pair); 22:33
camelia ===SORRY!=== Error while compiling <tmp>
Shape declaration with () is reserved;
please use whitespace if you meant a subsignature for unpacking,
or use the :() form if you meant to add signature info to the function's type
at <tmp>:1…
rf m: my $pair = :abc(123); -> :$key ($value) { say $key, $value; }($pair); 22:34
camelia Too few positionals passed to '<anon>'; expected 1 argument but got 0 in sub-signature of parameter $key
in block <unit> at <tmp> line 1
rf m: my $pair = :abc(123); -> ($value => $key) { say $key, $value; }($pair);
camelia ===SORRY!=== Error while compiling <tmp>
Preceding context expects a term, but found infix > instead.
at <tmp>:1
------> my $pair = :abc(123); -> ($value =>⏏ $key) { say $key, $value; }($pair);
Nemokosch I think you can, the syntax didn't click 22:37
:$key, :$value
m: my $pair = :abc(123); -> :$key, :$value { say $key, $value; }($pair); 22:38
Raku eval Exit code: 1 Too many positionals passed; expected 0 arguments but got 1 in block <unit> at main.raku line 1
Nemokosch yeah yeah
m: m: my $pair = :abc(123); -> (:$key, :$value) { say $key, $value; }($pair);
Raku eval abc123
rf Okay cool, thanks! 22:39
22:55 ispyhumanfly joined 22:59 ispyhumanfly left 23:10 freeside left
Geth ¦ doc: coke self-assigned Move HTML/CSS/js out of the repo github.com/Raku/doc/issues/2931 23:22
¦ doc: coke assigned to dontlaugh Issue The binary-based CI test is broken github.com/Raku/doc/issues/2859 23:26
¦ doc: coke self-assigned What's really autothreading and is it used correctly? github.com/Raku/doc/issues/2779 23:30
¦ doc: coke self-unassigned What's really autothreading and is it used correctly? github.com/Raku/doc/issues/2779
¦ doc: coke self-unassigned Spin off Pod::Convenience github.com/Raku/doc/issues/2696 23:32
23:35 ab5tract left 23:39 freeside joined 23:44 freeside left 23:45 nachtvlinder joined 23:48 nachtvlinder left 23:51 Nemokosch joined 23:59 freeside joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!