wayland Anyone know why I'd be getting a lot of: raku -Ilib -c lib/Qwiratry/QueryIterator.rakumod 00:03
Sorry, wrong message: Missing serialize REPR function for REPR MVMContext (BOOTContext)
That one. Anyone know why I'd be getting that? 00:04
It happens when running: raku -Ilib -c lib/Qwiratry/QueryIterator.rakumod
It happens when running: raku -Ilib -e 'use Qwiratry::Context; '
(sorry, wrong paste again) 00:05
...and the contents of that file are: role Qwiratry::Context is export {}
ugexe maybe a precomp error 00:09
try deleting lib/.precomp, or using -I. instead of -Ilib
wayland OK, thanks! 00:13
Neither seemed to help, unfortunately. 00:15
Unsetting RAKUDO_RAKUAST seems to have fixed it. But since I'm hoping to use RakuAST in this project, that's a problem. 00:17
I'll have a look at the Rakudo bugs, and see how I go.
grondilu MoarVM panic: Internal error: invalid thread ID 1575425 in GC work pass 00:20
^this happened in the middle of a fairly long computation (processing lots of chess grandmaster games) 00:21
it's going to be difficult to reproduce but I'll try
00:35 wayland left 01:23 wayland76 joined 01:25 sibl joined 01:34 sibl left 01:35 mahlay left 01:38 sibl joined 01:42 sibl left 01:49 sibl joined
Voldenet this: github.com/MoarVM/MoarVM/blob/main...ect.c#L435 02:26
given pretty wild thread id, I'm assuming it was uninitialized 02:29
grondilu: is this "invalid thread ID" error on x86 or arm? 02:30
(could be probably relevant because of memory ordering)
02:39 hulk joined, kylese left
grondilu x86 03:04
the weird thing is that this program runs fine for hours, but it did crash at least once. 03:05
for what it's worth it's a program that process PGN files on pgnmentor.com and turns them into polyglot files. 03:06
again, very hard to reproduce as it's running fine rn.
03:15 hulk left, kylese joined
Voldenet it could always be random bitflip because of no ECC, happens to everyone ;/ 03:50
04:05 lue joined 04:06 ShimmerFairy left 04:11 mahlay joined 04:16 mahlay left 04:19 lue is now known as ShimmerFairy 04:22 stanrifkin_ joined 04:23 lichtkind joined 04:24 stanrifkin left 04:25 lichtkind_ left 04:28 mahlay joined 05:01 wayland76 left 05:04 arkiuat left 05:20 arkiuat joined
disbot6 <jubilatious1_98524> Ahem... corrected answer now posted. 06:25
<jubilatious1_98524> weekly: unix.stackexchange.com/questions/8...066#803066 06:26
06:38 abraxxa joined 06:46 Aedil joined 06:57 silug left 06:59 eseyman left 07:00 manu_ joined, manu_ is now known as eseyman 07:09 silug joined
ab5tract wayland: is the code for Qwiratry::Context available for perusal? 07:39
07:52 Aedil left 07:57 stanrifkin_ left 08:12 arkiuat left 08:37 abraxxa left 08:41 arkiuat joined 08:46 arkiuat left 09:00 arkiuat joined 09:04 Sgeo left 09:05 arkiuat left 09:33 arkiuat joined, Aedil joined 09:37 arkiuat left 09:55 sibl left 10:11 arkiuat joined
disbot6 <simon_sibl> Is there any chance to see Raku support in cloudflare through wasm or something else in the near future ? 10:15
10:15 arkiuat left
Voldenet You could compile moarvm to wasm 10:31
10:33 sibl joined 10:37 sibl left 10:43 arkiuat joined 10:47 arkiuat left
lizmat feels inspired by raku-advent.blog/2025/12/19/day-19...gadungeon/ 10:49
10:53 eseyman left 10:55 manu_ joined, manu_ is now known as eseyman
lizmat simon_sibl not sure what you mean with "Raku support in cloudflare", could you elaborate ? 10:57
11:13 arkiuat joined 11:18 arkiuat left
grondilu has someone ever mananged to compile moarvm to wasm? 11:27
timo grondilu: that requires a port of libuv to wasm, or replacing the parts of libuv that moarvm uses with stuff that works on a wasm runtime 11:32
ab5tract is very pleased that the article resonated 11:44
11:48 arkiuat joined 11:54 arkiuat left 11:56 arkiuat joined 12:01 arkiuat left 12:29 arkiuat joined 12:34 arkiuat left 12:57 arkiuat joined
tbrowder tbrowder: i also feel grateful for it and all the other articles. the previous ones fit into the "gist" of the last one so neatly... 13:00
13:02 arkiuat left 13:31 arkiuat joined 13:36 arkiuat left 13:38 arkiuat joined, eseyman left 13:42 arkiuat left 13:58 arkiuat joined
disbot6 <antononcube> I added four new headline images in the Media folder of raku-advent.blog . 14:00
lizmat antononcube++ 14:04
14:11 arkiuat left 14:14 arkiuat joined 14:19 shareable6 left 14:21 shareable6 joined, arkiuat left 14:35 sibl joined 14:43 arkiuat joined 14:48 arkiuat left
disbot6 <simon_sibl> I meant for this: developers.cloudflare.com/workers/languages/ šŸ™ 14:59
15:06 arkiuat joined 15:11 arkiuat left 15:15 arkiuat joined 15:25 arkiuat left, Sgeo joined 15:36 arkiuat joined 16:19 human-blip left 16:21 human-blip joined 16:30 sibl left 16:31 librasteve_ left
disbot6 <aruniecrisps> I've got a silly question, is nqp going to continue to be something that people can use if they truly want to or will people be directed toward macros and RakuAST? 17:02
ugexe i don't think we direct people to use nqp
ShimmerFairy If you mean nqp:: ops, those were never really meant to be used outside of the Rakudo compiler; people only use them because of some deficiency in Raku or Rakudo itself. If you mean the program itself, then I don't see why you couldn't keep using it in the foreseeable future. 17:26
17:39 librasteve_ joined
disbot6 <aruniecrisps> The reason I'm asking is let's suppose someone wants to make a slang using nqp or they want to create a different HOW, would it still be available to them? Am I even asking the right questions? 17:46
17:55 manu_ joined
lizmat aruniescrips with raku.land/zef:lizmat/Slangify it should be almost always possible to not have to use NQP ops 18:35
if it is, then I would like to know about it :-)
disbot6 <aruniecrisps> lizmat: would the creation of a new form of ClassHOW like in OO::Actors or EnumHOW not count as using nqp ops or am I confusing something? 18:43
lizmat I don't see any NQP in OO::Actors ? 18:45
I see dependencies on the metamodel, yes
disbot6 <aruniecrisps> Ah okay I'm confused, thanks for the clarification 18:51
arkiuat I'm having some trouble using defined() on junctions. Here's some example code from docs.raku.org/type/Junction#method_defined 19:15
m: my $foo=Failure.new; say (one 3, $foo).defined;
camelia False
arkiuat The doc says that is supposed to say True, not False
and this is not a problem with failure: 19:16
m: my $bar; say (one 3, $bar).defined; say $bar
camelia False
(Any)
arkiuat that should say True also, but it doesn't
[Coke] do you mean (one 3, $bar) or one(3, $bar) ?
arkiuat oh, so it may be a typo in the example code, then? 19:17
m: say (one (3, $bar)).defined 19:18
camelia ===SORRY!=== Error while compiling <tmp>
Variable '$bar' is not declared. Perhaps you forgot a 'sub' if this was
intended to be part of a signature?
at <tmp>:1
------> say (one (3, <HERE>$bar)).defined
arkiuat m: my $bar; say (one (3, $bar)).defined
camelia False
arkiuat [Coke], apparently not
m: my $bar; say one(3, $bar).defined 19:20
camelia False
arkiuat To me this seems like a bug: the doc is correct but rakudo is misbehaving 19:21
I'd love to be wrong about that because I know how to fix the doc, but not how to fix rakudo
[Coke] m: say one(3).defined
camelia False
[Coke] m: say one(3.defined) 19:22
camelia one(True)
arkiuat I'm trying to check to see if exactly one member of a list of several things is defined (it's a failure condition otherwise), so that last case won't help me, sadly 19:23
19:23 mehbark left, lue joined, mehbark9 joined 19:26 ShimmerFairy left
ugexe arkuiat: if you want to take a crack at fixing it in rakudo i think you only need to look at this function and can probably figure it out from there github.com/rakudo/rakudo/blob/1592...#L114-L122 19:26
arkiuat ugh, a triple chained conditional operator 19:27
and defined-one is the final default? weird 19:28
ugexe github.com/rakudo/rakudo/commit/7a...dfc353a92e
that commit is the last time it was touched. you could potentially check if the old code did the right thing 19:29
arkiuat hmm, okay 19:30
looks like I'd need to fall back to 2020.01? 19:31
ugexe you could. sometimes i just copy paste the code into its own file and adjust things to test it that way
if you want to know how to actually go about trying your changes out here is what I do. I use rakubrew to build some version of raku, then i cd into ~/.rakubrew/versions/whatever. then i edit whatever rakudo source files and run `make test && make install` until it works
arkiuat okay, thanks! 19:32
let me just make sure I'm installing the correct ancient version first
ugexe you could also probably use monkey patching
arkiuat rakubrew is building me a 2020.01 19:35
build failed, message here: gist.github.com/arkiuat/52ca1e0c2b...a14347b496 19:39
I guess what I really ought to do at this point is file an issue, no? 19:42
I'm going to step away from the keyboard for a bit but will be back within half an hour, probably
19:49 librasteve_ left
ugexe perl6 -e 'say one(1, my $f).defined' 19:53
True
thats on 2018.01, so it used to work at some point
arkiuat I'll mention that in the issue, thanks 20:10
20:10 inspork left
arkiuat Math::Precision was on hold while I make a few PRs to Math::Angle for some additional units that I'd like it to have, but the first failure check in Math::Angle's TWEAK submethod is horrible and obviously should be rewritten as a one-junction... except that one-junction tests for definedness apparently haven't worked since before Math::Angle was first released three years ago 20:12
ugexe Here, the loop increments through eigenstates and continues only while the current value is NOT defined (isfalse(.defined)). It also requires nqp::not_i($seen++) to be true (i.e., seen is 0). However, this means it will stop before ever seeing a defined element, so seen never increments on seeing a defined valueā€”which is wrong for the "one" junction semantics. 20:29
that is what copilot on github said 20:30
gist.github.com/ugexe/dfbb0087a641...c019efc77e this was its suggested fix
"The bug occurs because the loop skips over defined values (since it continues only when .defined is false), meaning the seen counter does not increment for defined elements at all. As a result, it will always return 0 seen and thus iseq_i($seen,1) is false ($seen remains 0), so .defined returns False wrongly." 20:31
arkiuat I was checking to see if any related issues were already open, but I can't find any, or even a similar closed one 20:34
I'm putting a timestamped link to this IRC log, along with the examples that [Coke] and I tried already 20:35
github.com/rakudo/rakudo/issues/6048 20:43
okay, now I'm going to try to find some *other* clever way to rewrite Math::Angle's TWEAK 20:50
for now I'm using a a one-junction as a simple boolen and ANDing with a one-junction test for equality to zero. Any better suggestions? 21:20
that's with an `unless` in front of it 21:21
[Coke] bisectable6: say one(1, my $f).defined 21:23
bisectable6 [Coke], Will bisect the whole range automagically because no endpoints were provided, hang tight
arkiuat at least .defined still works on any-junctions :D 21:24
bisectable6 [Coke], Output on all releases: gist.github.com/d768b18edd39b58f23...1303d2360a
[Coke], Bisecting by output (old=2020.01 new=2020.02.1) because on both starting points the exit code is 0
[Coke] wow, broke nearly 6 years ago. 21:25
arkiuat yeah, I know, right! yikes.
here I was wondering why Math::Angle's TWEAK's failure check was written so weird, and instead now I'm wondering why the author didn't report the issue back then 21:26
or maybe they just don't use junctions. I haven't turn across any of them anywhere else in Math::Angle so far
ugexe i'd guess that junctions are going to be slower than not using them 21:34
arkiuat good point! maybe I'd better take that into account before sending in my PR 21:35
bisectable6 [Coke], bisect log: gist.github.com/c8047db7878c1b69ec...a822bd2500 21:41
[Coke], Output on all releases and bisected commits: gist.github.com/85d0b318cb9efcd701...af6db998b2
arkiuat [Coke]++ thanks, I've added those to a comment on the issue 21:47
grepping on *.defined and counting the elems that result also works for what TWEAK is trying to check here, and I think it's prettier than what I found. Shouldn't be any slower either 21:49
22:21 Aedil left 22:31 wayland76 joined
wayland76 .tell ab5tract Certainly! It can be seen at github.com/rakudo/rakudo/issues/6047 . I'm suspecting that this is something that's only happening on my computer (only because I think it's so obvious it would've been picked up by now), but it's definitely hampering me. 22:34
22:50 arkiuat left 22:53 arkiuat joined
grondilu m: print { .&exp / (1 + .&exp) }(10) 23:07
camelia 0.9999546021312976
grondilu m: print { .&exp / (1 + .&exp) }(10000)
camelia NaN
wayland76 If someone could reproduce it to confirm whether it's just me, that'd be good too.
grondilu m: print { .&exp / (1 + .&exp) }(1000) # kind of annoying but I understand raku is not a CAS 23:08
camelia NaN
[Coke] wayland76: works fine here on 2025.11 on my mac 23:13
tried multiple times with and without RAKUDO_RAKUAST=1
wayland76 [Coke]: Did you delete .precomp each time?
[Coke] Nope. Just tried again after deleting lib/.precomp, no change 23:15
just always passes. 23:16
grondilu: what's a CAS? 23:17
wayland76 [Coke]: OK, so sounds like it's just on my machine. That's inconvenient. 23:26
grondilu [Coke]: Computer Algebra System 23:31
I do wonder if most computer language chug like this though 23:32
perl: $_=1000; print exp($_)/(1+exp($_)) 23:34
no perl bot here?
p5: $_=1000; print exp($_)/(1+exp($_))
wayland76 [Coke]: How do I get the sw_vers output? Do I just run that at the Raku prompt? Because that doesn't work for me. 23:36

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!