🦋 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.
[Coke] just tried to build HEAD with ast enabled, got: 00:11
Class X::Redeclaration+{X::Comp} cannot inherit from X::Redeclaration because the parent is not composed yet
Geth rakudo/main: 2c9ad45c40 | ab5tract++ | 2 files
Make MetaPostfix::Hyper a WhateverApplicable

This fixes R#5901 (#5901). Pretty exciting that the WhateverApplicable implementation is robust enough that no real special treatment is required here.
00:48
linkable6 R#5901 [open]: github.com/rakudo/rakudo/issues/5901 RakuAST: `*>>.foo` producing incorrect (Q)AST
ab5tract I'm pretty sure that there is no fallout from this commit, but I get confused as to what test files are still expected to fail 00:51
07:34 donaldh- left
nine [Coke]: that's expected. You need to revert 3dad1a55ae1587cee58df8e10474a1551cf97741 to build with RAKUDO_RAKUAST=1 08:01
linkable6 (2025-06-07) github.com/rakudo/rakudo/commit/3dad1a55ae Revert "RakuAST: remove workaround needed by the old compiler frontend"
08:20 finanalyst left
nine MasterDuke: any progress on the RakuAST fix? 09:13
tellable6 nine, I'll pass your message to MasterDuke
09:40 donaldh joined
Geth rakudo/main: f2ffa8fbf3 | (Elizabeth Mattijsen)++ | lib/RakuAST/L10N.rakumod
RakuAST: actually handle localization of core enums

This moves already added core enums from the "term-" section to a new "enum-" section, and adds support for the Endian and SeekType enums.
11:31
rakudo/main: 7866988179 | (Elizabeth Mattijsen)++ | 3 files
RakuAST: add RakuAST::Enum class for handling enums

Intended to be used for core enums, but could also be used for custom ones. Is in fact a very shallow subclass of ::Term::Name, but allows for better deparsing and better readability.
Also added a "from-identifier" instantiator method to ::Term::Name to make it easier to create ::Term::Name objects that have simple identifiers, such as "True" and "False"
rakudo/main: f30501c481 | (Elizabeth Mattijsen)++ | 4 files
RakuAST: final fixes for handling enums specifically

Add enum- handling to highlighting, .raku and deparsing
EN/main: 284100653b | (Elizabeth Mattijsen)++ | 5 files
0.0.3
11:49
nine lizmat: I'd have called it RakuAST::Term::Enum to differentiate it more clearly from RakuAST::Type::Enum 11:57
lizmat ok, will do
I was in doubt about ::Enum or ::Term::Enum already :-) 12:02
Geth rakudo/main: 92068d9b46 | (Elizabeth Mattijsen)++ | 4 files
RakuAST: rename RakuAST::Enum to RakuAST::Term::Enum

to more clearly distinguish it from RakuAST::Type::Enum
12:08
DE/main: e5be650e09 | (Elizabeth Mattijsen)++ | 17 files
Initial commit after splitting off of L10N repo
12:20
DE/main: a7f387f68f | (Elizabeth Mattijsen)++ | 4 files
CI test prior to release
12:23
DE/main: dc781c1300 | (Elizabeth Mattijsen)++ | 2 files
0.0.2
12:25
FR/main: 5792cf35e9 | (Elizabeth Mattijsen)++ | 17 files
Initial commit after splitting off of L10N repo
12:47
FR/main: 8b78003545 | (Elizabeth Mattijsen)++ | 7 files
CI test prior to release
FR/main: 0cd5c2470f | (Elizabeth Mattijsen)++ | 2 files
0.0.2
12:50
HU/main: 1030c8302a | (Elizabeth Mattijsen)++ | 17 files
Initial commit after splitting off of L10N repo
13:06
HU/main: 8679b01273 | (Elizabeth Mattijsen)++ | 2 files
0.0.2
13:07
timo m: say "don't stop me now" 13:21
camelia don't stop me now
timo single quotes are of the devil 13:39
> ++ READLINE_LINE='raku -e '\''say "don'\''\'\'''\''t stop me now"'\'''
Geth IT/main: 770054ab85 | (Elizabeth Mattijsen)++ | 17 files
Initial commit after splitting off of L10N repo
13:49
IT/main: 7e19600f52 | (Elizabeth Mattijsen)++ | 2 files
0.0.2
13:52
lizmat hmm... whyis it sayin DE here? 13:53
ah, backscroll issue *phew*
Geth PT/main: 9226bfe1f7 | (Elizabeth Mattijsen)++ | 17 files
Initial commit after splitting off of L10N repo
14:06
PT/main: 4b38c25cb4 | (Elizabeth Mattijsen)++ | 2 files
0.0.2
14:08
jdv at this recent summit, which i didn't even know about, was there anything discussed regarding parsing perf? 14:35
nine don't think so 14:36
jdv or is that just "part of the post rakuast" stuff for a way later time kinda thing?
ok 14:37
[Coke] github.com/rakudo/rakudo/issues/5905 14:39
"Get core setting compile performance to an acceptable level"
jdv i read that as just getting rakuast close(r) to legacy 14:40
Geth L10N/main: 4cb625cfcb | (Elizabeth Mattijsen)++ | 89 files
CI test before release
15:01
L10N/main: c2595357b2 | (Elizabeth Mattijsen)++ | 5 files
0.0.3
15:14
[Coke] So, let's double check with a wider audience here: My take is we can't switch to RAKUAST if the performance goes down when we do it. 15:29
(primary blocker right now is still getting the install working, as I understand) 15:30
(not performance blocker, just in general)
jdv maybe it should be renamed RAKUFAST 15:32
lizmat [Coke]: agree, but with s/goes down/goes down too much/ at which I think slowdown of 20% would be acceptable for a first release 15:53
[Coke] I think from a marketing standpoint that'd be ungood. 15:55
lizmat the thing is, that without a release, people won't actually start using it extensively 15:56
perhaps we should create a separate release with RakuAST soon, so it would be easy for people to use it (once we get make install type issues fixed, clearly) 15:57
[Coke] folks, I'll be at TRPC this month and am happy to try to get a lightning talk in about Raku (probably 6.e related items). Please feel free to throw any potential topics at me. 15:59
like a pre-release? Sure, that works.
then especially need to fix install. :) 16:00
lizmat yeah, that'd be a prerequisite 16:14
nine To be honest, I find all the talk about marketing or release related to RakuAST kinda moot. It's been almost 2 1/2 months since I concluded my RakuAST work and in this time only a single one of the 3 issues I've found in make install has been fixed. And that was the trivial 2 lines fix. 16:25
And that's just make install. We're not even talking yet about zef or blin. 16:26
Geth ¦ problem-solving: lizmat self-assigned Towards a Raku Foundation github.com/Raku/problem-solving/issues/477
nine Which frankly, I don't understand. People are clearly interested in RakuAST. And there has been work around RakuAST. Even a recent bug report about something in RakuAST has quickly seen a fix. 16:27
So why is there so little work on the main path to an eventual release? 16:28
I could understand people just not being interested. Or people not having the time. Or people being too intimidated to try to fix anything. But none of these seem to apply :) 16:30
Note that this should in no way belittle the efforts that have been made. I'm super happy about the work that MasterDuke did at the RCS and the preliminary investigations into compiler performance. 16:37
[Coke] adds github.com/rakudo/rakudo/issues/5906 16:41
I am surprised there wasn't a ticket for it.
lizmat hmmm... that's a new failure mode
nine That's github.com/rakudo/rakudo/issues/5854 which is what MasterDuke++ is working on 16:42
[Coke]: there's no overall ticket. I created 3 tickets for the individual problems I've found instead
lizmat they can be made sub-issues :-) 16:43
[Coke] added as sub issue - thanks, nine. Was looking specifically for 'make install' in the text. 16:45
japhb nine: In my particular case, I have bandwidth for just one major Rakudo-improvement project, so I'm working on one that I know a bit about already and that no one has touched in 2 years. 17:46
17:56 El_Che joined
El_Che hallo, sadly rakudo 2025.05 still has the same problem with mimalloc of 2025.04: gist.github.com/nxadm/8cd284952966...6601e23b10 17:57
it breaks specifically alpine builds, I suspect the most popular container distro 17:59
timo i think the commit that checked out the specific version of mimalloc that didn't have the problem was backed out right after the release? did we not put it back for last month's release? 18:01
El_Che I am no aware of specific checkouts or rollbacks. When I tried to do it myself, it broke all over the place in my builds 18:07
(I mean in my local copy)
if there is no know security problem with the old one, maybe revert to it and bump it when it passes all tests? 18:08
[Coke] it was backed out and then someone added it back in 18:38
in their defense, it looked like the backout was accidental. 18:45
19:32 nine left, nine joined 21:10 mandrillone joined 21:16 mandrillone left