00:05 arkiuat joined 00:13 arkiuat left
Voldenet it's not actually impossible to compile raku into binaries - in fact given some specific AST it's possible to write C directly from it 00:16
generating C from tiny subset of the language should be trivial due to having built-in parser for it 00:19
for example "all types must be defined, only Str and Int are supported, only simple subs" should be relatively easy, types for scopes, signatures would need to be practically implemented in C 00:24
otoh, scopes, dynamic variables, await can't be written in trivial C, so that'd require bundling the whole raku runtime - it's not impossible to generate bytecode and use this path 00:27
00:27 arkiuat joined
Voldenet at some point I was planning to build some poc of embeddable raku (something like PerlInterpreter), it would definitely make this more feasible 00:30
arkiuat some Str somewhere isn't getting initialized. I think this is the kind of thing that made me flee *brew and scurry back into MacPorts (which hasn't updated rakudo since v2025.05) 00:33
Voldenet hm, is there any stack trace? 00:34
01:03 notable6__ joined, quotable6 left, notable6 left, bisectable6 left, linkable6 left, Ekho left, cryosis left, unicodable6 left, sourceable6__ left, coverable6 left, benchable6 left, releasable6 left, quotable6 joined, bisectable6 joined, linkable6 joined 01:04 nativecallable6_ left 01:05 nativecallable6 joined, zetaaaa_ left, Sgeo__ joined 01:06 cryosis joined, Ekho joined 01:07 benchable6 joined 01:08 Sgeo_ left
arkiuat Nope, it's a compilation error 01:08
pointing to /private/var/folders/p2/ with what look like hashes in the rest of the path 01:09
It happened most recently when I was installing ValueType and then later with App::Ecosystems 01:11
just now I tried "zef install rakudoc" -- same compilation error 01:12
perhaps when I decided to try *brew again, I needed to thoroughly eradicate MacPorts from my system first. Not sure that I've done that correctly 01:14
okay, this is fun. I just used rakubrew to switch from v2025.08 back to v2025.05, and now ValueType just installed with no problem, and rakudoc is installing 01:18
installed successfully, and now installing App::Ecosystems. So this seems to be a bug introduced in v2025.08 maybe? 01:19
01:21 hulk joined, kylese left 01:22 zetaaaa_ joined
arkiuat that failed, but for a much more graspable reason: expected module eigenstates v0.0.11 but got eigenstates v0.0.12 01:22
01:29 nine left 01:30 nine joined
Voldenet hmm, ValueType doesn't have very large test, maybe it could be further reduced to one line 01:32
arkiuat yes, that's a good tip 01:59
the compilation error singled out line 2 of the single test file, which is "use Test" which would certainly explain why this is happening across all modules! 02:04
however, "raku -MTest" under v2025.08 starts up with no problem 02:06
02:15 hulk left, kylese joined 02:49 zetaaaa_ left 03:08 arkiuat left 03:20 arkiuat joined 03:24 arkiuat left 03:25 dg left 03:26 dg joined 03:44 arkiuat joined 03:54 arkiuat left 04:36 Aedil joined 05:22 phogg left 05:27 phogg joined 06:26 Geth left 06:27 Sgeo__ left 06:28 kylese left 06:29 kylese joined 06:45 quotable6 left, cryosis left, quotable6 joined, bisectable6 left 06:57 dg left, timo2 left, timo2 joined
disbot9 <simon_sibl> I am not sure to understand the do whenever at docs.raku.org/type/IO/Socket/Async/ListenSocket why is the do required, why is the return value handled that way ? 06:58
07:00 dg joined 07:10 tadzik joined 07:13 merpp left, merp joined
disbot9 <simon_sibl> also, will Raku recognize such instruction and use sendfile or similar trick: $socket.write($path.IO.slurp(:bin)) ? or it will slurp the whole file in memory and only then send it through the socket ? 07:48
<simon_sibl> or linked to that question, would it be possible to lazily read a file, pipe it through a base64 generator and lazily send that output through a socket ? 07:50
08:09 dakkar joined 08:44 Maylay left 08:46 lichtkind_ joined 08:47 Maylay joined 09:23 mimosa_ joined 09:32 mimosa_ left 09:33 mimosa joined
ab5tract .IO.lines is lazy already, IIRC 10:15
`do` is a way of evaluating an expression into its sunk value. so here it's a sort of conditional `await` 10:16
FWIW, I've never seen that in the wild and I think it fails a docs smell test of only demonstrating as few language features at a time in an example 10:17
lizmat simon_sibl afaik, sendfile is out of reach at the moment: but I have set up irclogs.raku.org to be able to use that as soon as it becomes possible 10:19
although I think it will more become something like: $socket.sendile($path) 10:20
*sendfile
10:31 ds7832 joined 11:12 Geth joined 12:32 apa_c joined 13:27 ds7823 joined 13:30 ds7823 left, ds7832 left 13:31 ds7832 joined 13:38 ds7832 left 13:39 ds7823 joined 13:41 ds7823 left 13:43 ds7823 joined
disbot9 <jubilatious1_98524> I don't understand that do either. Usually I see do before a for loop, not within a block. 14:10
14:12 Sgeo joined
disbot9 <librasteve> Technically, do is a loop which runs exactly one iteration. .oO 14:15
<librasteve> [I have used do a bit but never really understood what it does]
14:15 ds7823 left, apa_c left
disbot9 <jubilatious1_98524> Don't we have once for that? 14:16
<librasteve> I am quoting the docs - docs.raku.org/language/control#do ... doesn't mean that I have any clue what they mean!
<jubilatious1_98524> I've used do to send the return of a for loop into a LHS assignment, usually overwriting the same object I started from. 14:18
14:21 arkiuat joined
ab5tract I hadn't ever considered `do` a looping construct. otherwise `do for` wouldn't make sense 14:25
once is per-scope. It can be used in a loop or outside of one 14:26
`do` is a statement prefix modifier that turns an expression into a value
disbot9 <jubilatious1_98524> m:my @a = "a".."e"; @a = for @a { $_.uc }; say @a;
<Raku eval> Exit code: 1 ===SORRY!=== Word 'for' interpreted as a listop; please use 'do for' to introduce the statement control word at /home/glot/main.raku:1 ------> my @a = "a".."e"; @a = for⏏ @a { $_.uc }; say @a; Unexpected block in infix position (two terms in a row) at /home/glot/main.raku:1 ------> my @a = "a".."e"; @a = for @a⏏ { $_.uc }; say @a;
ab5tract m:my @a = "a".."e"; @a = do for @a { $_.uc }; say @a; 14:27
evalable6 [A B C D E]
disbot9 <jubilatious1_98524> m: my @a = "a".."e"; @a = do for @a { $_.uc }; say @a;
<Raku eval> [A B C D E]
<librasteve> yeah - I read that do for returns the value of the last iteration (better go review what my do fors are doing!)
ab5tract sorry if I got that impression. I was referring to the final value in the statement list
arkiuat I've always thought of do as "what you use when you want to obtain the return value of a control statement usually evaluated in sink context" 14:28
ab5tract but it's probably not a good way to explain it
arkiuat: well said 14:29
arkiuat as for my zef install issues, I've had to fall back from v2025.08 to v2025.05 because at least I can install modules on it
disbot9 <librasteve> I am having a few niggles with installs atm 14:30
<librasteve> but does seem to be around 2025.08 / 2025.05 boundary ... 14:31
librasteve_ m: say do for ^3 {$++} 14:47
camelia (0 1 2)
librasteve_ m: say for ^3 {$++} 14:48
camelia ===SORRY!=== Error while compiling <tmp>
Unsupported use of bare "say". In Raku please use: .say if you meant to
call it as a method on $_, or use an explicit invocant or argument, or
use &say to refer to the function as a noun.
at <tmp>:1
-…
disbot9 <librasteve> yeah - that's what I expected do for to do (ie do the for loop and return the results as a List) 14:49
<librasteve> weekly: unix.stackexchange.com/questions/3...463#800463 14:51
<librasteve> lools like the weekly bot is down here (Discord) 14:52
librasteve_ weekly: unix.stackexchange.com/questions/3...463#800463
notable6__ librasteve_, Noted! (weekly)
librasteve_ but not here (IRC)
arkiuat I've been using do + given a lot lately 14:53
do + given-when, I should say 14:54
14:55 apa_c joined
arkiuat ab5tract, thanks! 14:56
and by a lot, I mean two or three times, which only seems like a lot because it's a new usage for me 14:57
14:57 guifa joined 15:01 apa_c left
disbot9 <librasteve> yeah it took me a while to get into it ... such as 15:02
<librasteve> my @todos = do for <one two> -> $text { Todo.new: :$text };
arkiuat librasteve, that stackexchange example has me wondering if the one-and-only distinction between for and given is that given doesn't loop 15:03
15:03 abraxxa left
disbot9 <librasteve> that's how it feels to me 15:03
lizmat m: .say for 42 15:04
camelia 42
arkiuat I don't think I've ever used for-when yet
lizmat m: .say given 42
camelia 42
lizmat m: .say for (1,2,3)
camelia 1
2
3
lizmat m: .say given (1,2,3)
camelia (1 2 3)
disbot9 <librasteve> I think the only thing given does is to set the topic afaik
lizmat indeed 15:05
arkiuat lizmat, oh, very interesting
lizmat and it will not iterate on iterables
arkiuat so that's a second difference, thanks
disbot9 <librasteve> .oO (gee I love this language!)
arkiuat and that totally makes sense, given the close connection between looping and wanting to iterate over iterables vs treating an iterable as an item 15:06
that makes the distinction between for and given a lot clearer for me 15:07
librasteve, so I guess a series of whens followed by a default (with scattered proceeds and succeeds as needed) would work inside any topic-setting control syntax? 15:09
lizmat m: $_ = 42; when 42 { say "success" } 15:10
camelia success
arkiuat Nice.
disbot9 <librasteve> I bow to lizmat in all matters, natch 15:11
15:12 apa_c joined
arkiuat that's incredibly flexible 15:14
I am going to be on the lookout for opportunities to use for-when now! 15:17
15:18 apa_c left
arkiuat thanks, lizmat 15:21
disbot9 <jubilatious1_98524> @arkiuat I wrote it both ways, given/when and for/when, nd finally decided to post the for/when version. 15:54
arkiuat I'm glad you did!
disbot9 <jubilatious1_98524> m: $_ = 43; when 43 { say "success" }; 16:06
<Raku eval> success
<jubilatious1_98524> m: $_ = 43; when none 42 { say "failure" };
<Raku eval> failure
<jubilatious1_98524> Use none to negate. 16:07
arkiuat that's certainly more elegant-looking than when $_ ne 42, but I expect non-rakuuns are going to cite as yet another of Raku's weird idioms 16:16
sorry, meant !=, not ne
disbot9 <jubilatious1_98524> m: $_ = 43; when $_ != 42 { say "failure" } 16:23
<Raku eval> Exit code: 1 ===SORRY!=== Error while compiling /home/glot/main.raku Confused at /home/glot/main.raku:1 ------> $\⏏_ = 43; when $_ != 42 { say "failure" }
<jubilatious1_98524> m: $_ = 43; when $_ != 42 { say "failure" } 16:24
<Raku eval> Exit code: 1 ===SORRY!=== Error while compiling /home/glot/main.raku Confused at /home/glot/main.raku:1 ------> $\⏏_ = 43; when $\_ != 42 { say "failure" }
<jubilatious1_98524> Bot interprets $ ... $ as italics (markdown). Anyway $_ != 42 works also. 16:25
16:29 dakkar left 17:00 arkiuat left 17:05 arkiuat joined
arkiuat yes, but none 42 looks prettier 17:13
disbot9 <librasteve> I recommend back ticks as in m: code when in discord 17:14
Voldenet btw, "none" isn't even the same thing 17:20
disbot9 <jubilatious1_98524> Thanks Steve.
Voldenet under the hood I mean, it's a lot more complicated than plain != 17:22
disbot9 <jubilatious1_98524> m: my @a = "a".."e"; @a.grep( {! /f/ } ).say; 17:23
<Raku eval> (a b c d e)
Voldenet m: given 42 { when * != 42 { say "failure" }}; # another way of writing it
camelia ( no output )
disbot9 <jubilatious1_98524> m: my @a = "a".."e"; @a.grep( none /f/ ).say;
<Raku eval> (a b c d e)
<jubilatious1_98524> m: my @a = "a".."e"; @a.grep( none /e/ ).say; 17:24
<Raku eval> (a b c d)
Voldenet m: sub x(Mu $x) { say $x }; x none 42; 17:25
camelia none(42)
Voldenet it's useful, but plain op for "not equals" can't do the above, for example 17:26
disbot9 <jubilatious1_98524> Why are you complicating things by creating a sub?
Voldenet Obviously to show that you can complicate things by complicating them ;) 17:27
disbot9 <jubilatious1_98524> [ BTW, grep examples are straight from the Docs ].
Voldenet huh, I've checked and apparently grep(none /f/) is faster 17:30
so apparently in some cases junctions are even faster than simple negation 17:32
I can guess that the `none /e/` doesn't even create a block and blocks aren't cheap
disbot9 <jubilatious1_98524> m: $ = 43; when $ eq none 42 { say "failure" }; 17:34
<Raku eval> failure
<jubilatious1_98524> The Junction page suggests: "In general it is more readable to use a positive comparison operator and a negated junction". 17:35
Voldenet hm, I remember there was some weird gotcha about negated junctions 17:38
m: say none([]) != 42 17:39
camelia False
Voldenet which means that the array contains value 42… 17:41
so I tend to avoid junctions, because I simply don't understand the why behind it 17:42
arkiuat well, not exactly. It means that none of the values are unequal to 42, and there are no values, so none of them are 17:43
neither != nor ne will accept Nil as an arg 17:44
Voldenet I understand technically, but I intuitively would use it wrong 17:45
arkiuat oh wait, I'm wrong. It complains with an error msg, but:
m: say so Nil != 42
camelia Use of Nil in numeric context
True
in block <unit> at <tmp> line 1
arkiuat it does return True. Huh.
Voldenet m: say so 42 (elem) [] # this is what I prefer 17:47
camelia False
arkiuat sure, that's much clearer
Voldenet and 17:48
m: say so 42 !(elem) []
camelia True
Voldenet otoh when there's no negation, it works as I'd expect 17:49
m: say none() eq 42 17:50
camelia none()
Voldenet m: say so none() eq 42
camelia True
arkiuat now I'm even more confused 17:57
Voldenet hehe, I know right? :) 17:59
the trick is in how negation is implemented
it simply does !($op1 eq $op2) 18:00
so in case of negated junctions… it actually finds equal values and negates the result
18:00 [Coke]_ joined 18:03 [Coke] left
lizmat Voldenet arkiuat github.com/Raku/problem-solving/issues/319 18:10
arkiuat lizmat, thanks! 18:14
oh no, and I just told tbrowder the other day that !(elem) is equiv to ∉ 18:19
I suppose in most context it still is
lizmat yeah, it is (usually)
arkiuat so it looks as if some of the confusing behavior we were discussing will change in 6.e 18:23
lizmat it does ? 18:35
arkiuat okay, that's surmising too much. there are some unmerged PRs outstanding that haven't been closed that *seem* to address at least part of the issue 18:38
disbot9 <jubilatious1_98524> @Voldenet are you equating say so none() eq 42 with say so $_ eq none 42?
arkiuat but the main one is still marked DRAFT, so definitely surmising too much
18:40 silug4 left 18:41 silug4 joined 19:28 guifa left 19:30 disbot10 joined 19:31 charsbar_ joined, dano2 joined 19:32 guifa joined, ab5tract_ joined, oodani joined, gabiruh joined 19:33 [Coke]_ is now known as [Coke] 19:34 greaser|q_ joined 19:35 sivoais_ joined, andinus2 joined 19:36 ab5tract left, camelia left, charsbar left, ab5tract_ is now known as ab5tract, ugexe left, oodani_ left, disbot9 left, andinus left, dano left, gabiruh_ left 19:37 greaser|q left, sivoais left, dano2 is now known as dano, charsbar_ is now known as charsbar 19:38 linkable6 left, greppable6__ left, committable6 left, benchable6 left, nativecallable6 left, notable6__ left, quotable6 left, bloatable6 left, evalable6 left 19:40 shareable6 joined, linkable6 joined, releasable6 joined, bloatable6 joined 19:41 committable6 joined, greppable6 joined, tellable6 joined, evalable6 joined, bisectable6 joined 19:42 coverable6 joined, sourceable6 joined 19:43 unicodable6 joined, benchable6 joined, nativecallable6 joined, notable6 joined, quotable6 joined 19:44 Aedil left 19:47 ugexe joined, jdv joined 19:51 camelia joined 20:00 camelia left 20:07 arkiuat left 20:08 apa_c joined 20:19 arkiuat joined 20:23 camelia joined 20:24 arkiuat left 20:38 stanrifkin joined 20:39 melezhik_ joined
melezhik_ . 20:39
20:39 arkiuat joined 20:41 itaipu left 20:47 melezhik_ left, melezhik_ joined
arkiuat seen on another channel: "rak will rename with `git mv' if it sees the file is under git control" <-- this is super cool 20:48
lizmat++
20:49 itaipu joined
Voldenet jubilatious1_98524: Yes, `none of elements () is equal to 42` and `from () none is equal to 42` sound very similar to me, but that only works with these specific operands 21:03
but it might be that I'm not seeing something, I'm not very fluent in junctions 21:05
I still prefer the code I know best from C#, so explicitly checking if element is in collection or if it matches some predicate 21:07
21:10 Manifest0 joined 21:11 ds7832 joined 21:15 lichtkind_ left 21:18 melezhik_ left 22:00 halloy2489 joined, arkiuat left 22:06 arkiuat joined 22:10 ds7823 joined 22:16 apa_c left 22:21 ds7832 left 22:38 halloy2489 left, Geth left 22:40 itaipu left 22:45 itaipu joined 22:49 ds7823 left 22:52 ds7832 joined
disbot10 <jubilatious1_98524> m: $ = 43; say so $ == none(42); 22:55
<Raku eval> True
<jubilatious1_98524> m: $ = 43; say $ ~~ none(42);
<Raku eval> True
23:16 librasteve_ left

Powered by Cro and the Raku® Programming Language.

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

Thank you!