🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
01:52 MasterDuke joined
MasterDuke lizmat: i think there's a new infini-loop. after pulling the latest nqp commits, the rakudo build hangs in `+++ Generating  gen/moar/BOOTSTRAP/v6c.nqp` and `perf top` shows moarvm spending *lots* of time in `get_attribute` 01:54
03:58 MasterDuke left 05:50 [Coke] left 05:53 [Coke] joined 06:45 djinni` left 06:50 djinni` joined
nine For me it hangs in nqp-m --module-path=blib --ll-exception --target=mbc --output=blib/Perl6/Metamodel.moarvm gen/moar/Metamodel.nqp 07:43
lizmat grrr 08:20
confirmed 08:21
Geth nqp/main: 57b956822e | (Elizabeth Mattijsen)++ | src/how/RoleToRoleApplier.nqp
Fix infinilooping in Rakudo

Clearly we need more tests in NQP :-(
nine lizmat: don't you run a spectest after changes in NQP? 08:49
lizmat: btw. why not return 1 instead of $skip := 1 in can_skip_attribute_by_name? If we find the attribute already in that list, we know it cannot be a duplicate anymore and further checks are unnecessary. 08:51
lizmat nine: I thought I had run a spectest, but evidently I didn't :-(
nine I know how that is :) 08:52
Or even worse, you do run a spectest but only did make and not make install in NQP, so you only think you tested your change
lizmat re returning 1: I just followed the original code, and was't sure about the semantics, as it *also* checks for name collisions 08:53
08:55 [Coke] left
lizmat nine: what I find strange in all this, is that actually cannot find a use of the NQP RoleToRoleApplier in rakudo 08:58
08:58 [Coke] joined
lizmat evidently I don't understand the bootstrap well enough yet :-( 08:58
08:59 epony joined
gfldex lizmat: Do you use git-hooks before `git push`? 09:03
lizmat no 09:05
gfldex I found those helpful to avoid embarrassment. :) 09:06
lizmat well, there's that, but I already run spectests at least 20 times / day, that's more than 1 hour I can spend on social media already :-) 09:09
09:18 finanalyst joined 09:25 epony left, epony joined 09:58 sena_kun joined 10:26 finanalyst left
Geth nqp/main: 5d896651d9 | (Elizabeth Mattijsen)++ | src/how/NQPClassHOW.nqp
Streamline NQPClassHOW some more

  - add some more helper subs to simplify updates
  - add "add_tweak" method to support role TWEAKS in the future
nqp/main: aadfaf9138 | (Elizabeth Mattijsen)++ | src/how/NQPConcreteRoleHOW.nqp
Streamline NQPConcreteRoleHOW

  - add some helper subs
  - add protected blocks where necessary (functional at a later stage)
  - add "add_tweak" method to allow adding multiple role tweaks
14:14 [Coke] left 14:19 [Coke] joined 16:48 nebuchadnezzar left 17:49 librasteve_ joined
librasteve_ o/ 17:53
I just made a minor PR on rakudo.org (this is a new version that addresses the comments of others - as you can see in the comments) — github.com/rakudo/rakudo.org/pull/100 17:55
would it help to ping someone to review and merge? if so - who?
(I am happy to be added to the rakudo.org admin if that’s a better option) 17:57
lizmat librasteve: I'm missing a reference to El_Che's packages, is that intentional ?
librasteve_ you mean rakudo-pkg (is El_Che nxadm?) 17:58
lizmat yes
librasteve_ this PR does not change the current ref to rakudo-pkg … I tested that this morning on a clean ubuntu build and it came up with 2022.10 … so I have moved this aspect out to a new Issue github.com/rakudo/rakudo.org/issues/99 17:59
and asked them to get it working first 18:00
lizmat ack... invite for commit bit on its way!
librasteve_ ta
jdv is installing a modern raku hard?
librasteve_ it’s easy if you know how 18:01
jdv i must of missed something in my backlogging
librasteve_ the hard part i navigating all the options
jdv there are quite a few different ways 18:02
this all came about because of user complaints? 18:05
librasteve_ yes - www.reddit.com/r/rakulang/comments...libraries/ 18:08
jdv ah 18:09
did we lose our deb porter?
librasteve_ standard apt-get install rakudo is waaay out of date, so I guess yes 18:10
jdv i mean deb is normally out more than others 18:11
but experimental is even super out
librasteve_ since the rakudo_pkg folks are keen, I think the best short term approach would be that (ie attach that cloudsmith repo to your apt-get) … so I am hoping that they will get that up and running (currently it loads 2022.10 for me) 18:12
jdv what is that now? i think the repo is newer...
librasteve_ 2024.01 18:13
lizmat librasteve_: could it be that your Linux is EOL ?
jdv iirc Blin uses that repo and it was up to date last i saw
librasteve_ ideally we would get the deb guys on board … but we can also remove some of the footguns on our own sites
let me double check my linux 18:14
lizmat librasteve_: if your Linux is EOL, then El_Che doesn't provide newer Rakudo versions anymore
librasteve_ Welcome to Ubuntu 22.04.1 LTS 18:16
jdv Blin uses deb stable at the moment. i only know about debian. 18:17
18:25 epony left
librasteve_ ah you have to guess its ‘apt-get install rakudo-pkg’ … let me clarify with bazzaar over on the issue 18:25
18:25 epony joined
Geth rakudo/lizmat-native-int-coercion: b3e846542a | (Elizabeth Mattijsen)++ | 2 files
Provide native int coercers to Int/Cool

  - .int, .int64, .int32, .int16, .int8
  - .uint, .uint64, .uint32, .uint16, .uint8, .byte
Note that this is just to provide the API. The implementation is ... (25 more lines)
rakudo: lizmat++ created pull request #5526:
Provide native int coercers to `Int`/`Cool`
nqp/main: 1a399b24cb | (Elizabeth Mattijsen)++ | src/how/RoleToClassApplier.nqp
Streamline RoleToClassApplier

  - add some helper subs
  - rename some variables for clarity
  - replace for loops by while loops
  - remove some dead code (was marked todo since 2011 :-)
19:33 [Coke]_ joined 19:35 [Coke] left 19:37 [Coke]_ is now known as [Coke]
lizmat meh, it might be that NQP has borked Rakudo build once again :-( even though the spectest was clean :-( 19:38
[Coke] someone has requested "copilot business" for the raku org on github 19:39
lizmat yeah, I saw that? cannot find out who did that though 19:41
How can I run a single test file when built with --with-nqp ? 19:42
./raku t/spec/S06-operator-overloading/imported-subs.t just says that it can't find Test :-( 19:43
vrurg To see who's requesting it Copilot needs to be activated first.
lizmat ./raku -Ilib t/spec/S06-operator-overloading/imported-subs.t fails with Missing or wrong version of dependency 'gen/moar/stage2/NQPHLL.nqp' 19:44
vrurg lizmat: try removing ~/.raku/precomp
lizmat no change :-( 19:45
vrurg There is something off about serialization currently. It feels like compiler's signature may not be changing something or something that causes re-use of older precomps.
lizmat there's *definitely* something off there :-( 19:46
vrurg Looks like only removal of the entire installation dir is on the table now. 19:47
lizmat yeah... :-( 19:49
even building with --gen-nqp=2024.01-32-g1a399b24c doesn't allow me to run the test file :-( 19:53
Ahhh I think I know why it first succeeded: if the libs that are used in that test file have already been precompiled, the tests pass 19:57
so it seems I borked precomping the libraries :-( 19:58
lizmat is going to contemplate further research
librasteve_ hi folks, on Rakudo Star, I have started a discussion issue 22:03
the gist is *Please do comment if you would like to restore Star to an effective option - I would love to hear the plan. Meantime, I propose we remove this option from the main rakudo.org menu.* 22:04
jdv if star was just rakudo+zef id be into it 22:05
last i looked it was a bunch
perhaps if it was a minimal "batteries included" but idk if thats possible... 22:06
gfldex I'm trying to find the origin on "Cannot unbox 65 bit wide bigint into native integer" to add a little awesome. Neither ack not a stacktrace are helpful. Any pointers? 22:07
librasteve_ jdv: the current journey for a newb is (i) raku.org, (ii) rakudo.org, (iii) rakudo star maybe … I think by pruning Star we can just give everyone rakudo + zef and cut a jump / decision out 22:09
jdv perhaps 22:10
librasteve_ why only perhaps?
what are your concerns? 22:11
jdv good luck getting agreement as to what star includes
otherwise, sounds great!
librasteve_ not needed, this proposal is just to prune the Star variants period 22:12
or if someone wants to keep it they need to step up and bring Star back up to date 22:13
2020.04 for macOS Intel only doesn’t help us 22:14
jdv cool
librasteve_ guess I should leave it a day or two to see if anyone wants to raise concerns 22:16
22:30 sena_kun left
[Coke] It may take a bit for folks to digest it. Good luck. (seriously, I'd like to see it go at this point) 22:34
22:36 epony left 22:48 MasterDuke joined
MasterDuke gfldex: github.com/MoarVM/MoarVM/blob/main...P6bigint.c for the 'Cannot unpack ...' error 22:50
gfldex Scary scary C-land. :->
MasterDuke [Coke]: btw, and speaking of error messages, i keep meaning to try and improve the error message in that particular case  of declaring a typed array and passing an array of the wrong type (and also of declaring an array of type Array) 22:54
23:35 MasterDuke left