Geth rakudo/main: 15 commits pushed by (Nick Logan)++
review: github.com/rakudo/rakudo/compare/6...8c8fd6a36e		 00:48
rakudo: ugexe++ created pull request #6673:
RakuAST: fix a tenth batch of deparse holes found by round tripping raku code		 00:50
rakudo/main: 10 commits pushed by (Nick Logan)++
review: github.com/rakudo/rakudo/compare/e...449aca396c		 03:51
rakudo: ugexe++ created pull request #6674:
RakuAST: fix an eleventh batch of deparse holes found by round tripping raku code		 03:57
07:10 djinni`_ joined, djinni` left
lizmat [Coke]: do we actually run a blin with RAKUDO_RAKUAST=1 RAKU_LANGUAGE_VERSION=v6.e.PREVIEW ? 10:01
that would have shown Needle::Compile to fail since github.com/rakudo/rakudo/commit/90...9a084ccd42
iow, I think we need such a blin run as it will make sure that modules without specific version requirements will continue to run on a language level increase 10:03
10:42 finanalyst left
Geth roast/remove-js-fudging: 212cb9c04b | (Elizabeth Mattijsen)++ | 40 files
Remove fudging for the obsolete Javascript backend
 11:52
roast: lizmat++ created pull request #913:
Remove fudging for the obsolete Javascript backend
rakudo/main: 5 commits pushed by (Elizabeth Mattijsen)++ 12:14
rakudo/main: 627decc752 | (Elizabeth Mattijsen)++ | lib/Test.rakumod
Reduce number of dynvar lookups when testing

In most cases from 6 dynvar lookups to 3. Makes spectest about 2% faster.		 13:13
ugexe If things arenā€™t marking their language level and break they already know how to fix their code: mark the language revision it uses 13:24
0.0.10 2026-05-25T19:58:30+02:00 13:27
- Fix code for post 6.e behaviour of %_
Literally from the change log of needle compile
lizmat I forgot to test with RAKUDO_RAKUAST=1 the fix was incomplete
I didn't fix the test 13:28
ugexe Itā€™s been passing blin runs with rakuast 13:33
lizmat ah, but not with RAKU_LANGUAGE_LEVEL=vt 13:34
v6.e.PREVIEW
ugexe Well the be clear I have enough other stuff Iā€™d want to get completed before Iā€™d bother fixing modules that havenā€™t pinned their revision in 29 years, before the release was supposed to happen. But maybe someone else can power through those 13:43
s/29/20/ 13:44
[Coke] thanks to ugexe's research, we have a 99% successful blin run. it's now hanging on 14:01
[2026-09-13T00:21:30-04:00] šŸ„žšŸ„ž Saving the dot file
which it's been hanging on for... 12+ hours? (it's done that before). but it didn't get OOM killed, which is the important part. :) 14:02
lizmat well, I would suggest doing a blin run with RAKUDO_RAKUAST=1 *and* RAKU_LANGUAGE_LEVEL=v6.e.PREVIEW 14:03
before my last fix to Needle::Compile, that *would* have turned up as a failure 14:04
I would think there may be more modules suffering from not having a 6.d or properly using %_ 14:07
which would otherwise only show up after we properly release 6.e
[Coke] ok. I have a few dozen more commits on main to get through, then will make a branch for that. 14:09
Can we set RAKU_LANGUAGE in src/main.nqp or is that too late? 14:10
lizmat RAKU_LANGUAGE_LEVEL is only read in the Raku grammar 14:11
but isn't it easier to set it in the blin environment ? I mean, that was the point of making it an env variable as well :-)
[Coke] the problem with doing it as an env var, is that var is set for OLD and NEW versions. 14:14
whereas the trick we were doing with RAKUAST and switching src/main.nqp, only impacts that one commit on the branch. 14:15
so we could compare "with and without", not "new and old version with"
lizmat the check is in Raku/Actions.nqp: nqp::getenvhash()<RAKU_LANGUAGE_VERSION> 14:17
m: use nqp; my $a := nqp::getenvhash; nqp::bindkey($a,"FOO",42); dd nqp::atkey($a,"FOO") 14:18
camelia 42
lizmat m: use nqp; nqp::bindkey(nqp::getenvhash,"FOO",42); dd nqp::atkey(nqp::getenvhash,"FOO") 14:19
camelia 42
lizmat looks like binding a key to nqp::getenvhash works
so something like:
[Coke] (doing a reduced redo run to weed out some potential false positives) 14:20
lizmat nqp::bindkey(nqp::getenvhash,"RAKU_LANGUAGE_LEVEL","v6.e.PREVIEW")
in main.nqp should do the trick
[Coke] ah, cool. OK.
When I get there, I'll branch off HEAD and add those two changes.
lizmat cool!
ugexe Well my point is just add the language revision to the code. If we are just gonna always fix everything for everyone then what is the point of the language revisions at all, letā€™s just make the breaking change 14:30
lizmat my point is discovery, so we can fix modules by adding a language level or other fix in a PR 14:32
this is not about reverting the %_ fix 14:33
Geth rakudo/main: 9 commits pushed by (Nick Logan)++ 14:36
[Coke] seeing about a dozen errors, but most like: 14:39
No candidates found matching identity: AttrX::Lazy^M
Geth rakudo: ugexe++ created pull request #6675:
RakuAST: fix a twelfth batch of deparse holes found by round tripping raku code
timo [Coke]: isn't ^M what you see when you have a dos-formatted file in a linux or mac system? wrong line endings? 14:40
[Coke] Looks like that's coming from systemd 14:41
but yes, that's when you normally see that 14:42
(but I see it on a file on a linux system that was generated on that same system)
timo curious. "from systemd"? if you see it in the journal, maybe looking at the same line with a different output, like --output=json-pretty can tell you more information? 14:44
or it could have something to do with the fact that you're running the thing with TTY so what it outputs is stuff like you would output to a TTY and not stuff you would output to a log file, so newlines may behave differently based on terminfo and termcap and whatever else 14:45
on a terminal in a specific mode, raw I assume, outputting a newline just puts the cursor one line down, you'd also output a carriage return to put the cursor on the left of the terminal, so it makes a lot of sense actually to see ^M in the output when it's taken straight from the process and put right into the journal 14:49
[Coke] The ^M is the *least* of my concerns right now. :) 14:53
Given I'm just coming off a week of the OOM killer being very mean to me.
when I say systemd - it's in the wrapper output before and after the process.
timo I still think we should pair on this topic 14:54
[Coke] so I see the test output for Module X - that has text before and after with info from the systemd runner.
It's not in the system logs. 14:55
s/in/from/
but yes, happy to track it down at some point! 14:58
timo I can't make heads nor tails from your text description unfortunately
for now, I'm due for a little nap
[Coke] github.com/Raku/Blin/blob/main/lib...akumod#L21 14:59
That's the call that invokes systemd-run
that is saved by blin and then dumped out if there's a failure in the output/ directory. 15:00
timo ok, that's what you call the systemd-wrapper
[Coke] I didn't save the one that had the ^M in it (trying to get the blin results on modules that didn't work on the last full run). When I get another one, I'll gist it 15:01
gist.github.com/coke/21f1eec0b1960...82235653be 15:07
the ^M's are on the *inside*, so they are being captured on the lines output by zef. 15:08
15:19 finanalyst joined 15:51 lizmat left 16:54 finanalyst_ joined, finanalyst left 16:59 finanalyst_ left, finanalyst joined 17:01 finanalyst left 17:02 finanalyst joined 17:04 finanalyst left 17:10 finanalyst joined
Geth roast: ugexe++ created pull request #914:
Put a leading declarator doc before the routine it documents		 17:37
roast: dd85d3c995 | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
Put a leading declarator doc before the routine it documents (#914)

A leading doc attaches to the declarator immediately after it, which for
  `#| doc` above `my $anon-sub = anon Str sub {}` is the variable, not
the sub in its initializer. These tests document the sub and the block, so the doc now precedes them inside the initializer.		 17:41
linkable6 ROAST#914 [closed]: github.com/Raku/roast/pull/914 Put a leading declarator doc before the routine it documents
Geth roast: ugexe++ created pull request #915:
Sync even more recent roast adjustments to 6.c-errata		 17:45
roast: ugexe++ created pull request #916:
Sync even more recent roast adjustments to 6.d-errata
[Coke] github.com/coke/raku-blin-release-results - I think we have two actual failures with the run up through 00fca760bb 17:49
linkable6 (2026-09-08) github.com/rakudo/rakudo/commit/00fca760bb Merge pull request #6657 from ugexe/ugexe/subst-sigspace-no-samemark
ugexe i already opened PRs for those github.com/MARTIMM/XmlActions/pull/4 github.com/alabamenhu/PolyglotRegexen/pull/2 17:50
Geth Blin/main: 2018084286 | (Will Coleda)++ | util/run-blin.sh
Update systemd runner options

Without the "omit", the blin process has been getting killed far too often.		 17:53
[Coke] good, thanks. 18:07
Geth rakudo/main: 8 commits pushed by (Nick Logan)++ 18:25
rakudo: ugexe++ created pull request #6676:
RakuAST: fix a thirteenth batch of deparse holes found by round tripping raku code		 18:27
rakudo: ugexe++ created pull request #6677:
RakuAST: drop a parameter's meta-object when its where or type changes		 18:48
rakudo: ugexe++ created pull request #6678:
RakuAST: report a worry on the line of its cursor
18:53 lizmat joined 20:02 lizmat left 20:05 lizmat joined
Geth rakudo/main: 7 commits pushed by (Nick Logan)++ 20:59
rakudo/main: f29f674415 | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
RakuAST: report a worry on the line of its cursor (#6678)

Previously a worry raised after the parse had gone past its cursor named the line the parse had reached while its excerpt showed the cursor: a
  "Missing declarand for trailing declarator doc" worry named the next
declaration or the end of the unit, and the TEMP phaser worry the line after its whitespace. build-exception moves a cursor the parse got past ... (5 more lines)		 21:00
lizmat I just noticed something weird: on a freshly installed rakudo, the startup time for me is 0.06 seconds 21:02
however, if I install my regular set of modules (such as App::Rak), the startup time goes up to 0.08 seconds 21:03
so: startup times appears to related to the number of modules installed? Even though I'm not loading any module?
it's not resolve-unknown-repos 21:07
Geth rakudo: ugexe++ created pull request #6679:
RakuAST: fix a fourteenth batch of deparse holes found by round tripping raku code		 21:08
rakudo: ugexe++ created pull request #6680:
RakuAST: check an attribute default against the base type of its meta-object		 21:09
21:20 finanalyst left
Geth nqp/remove-js-backend: 627937cf60 | (Elizabeth Mattijsen)++ | 113 files
Remove the Javascript backend support

See: github.com/Raku/problem-solving/issues/527		 21:23
nqp: lizmat++ created pull request #875:
Remove the Javascript backend support
ugexe i cant reproduce, the module loading env vars dont show anything being loaded but core, and trace output doesnt show any files being oepned 21:24
lizmat so you're not seeing a difference in bare startup time? 21:26
ugexe no 21:32
if you do like `mv share/perl6/site share/perl6/site.bak` does your startup time go back down? 21:33
Geth rakudo/main: bc78a3c7d1 | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
RakuAST: drop a parameter's meta-object when its where or type changes (#6677)

Previously set-where and set-type on a RakuAST::Parameter changed the tree and the check the binder compiles inline, but the Parameter object made at its BEGIN time kept the old constraint and type. A fresh signature and routine built around the mutated parameter then accepted what the new constraint accepts on the fast path and what the old one ... (9 more lines)		 22:26
rakudo/main: 5 commits pushed by (Nick Logan)++
22:48 shareable6 left, notable6 left, releasable6 left, bisectable6 left, sourceable6 left, bloatable6 left 22:50 nativecallable6 left, quotable6 left, unicodable6 left 22:51 releasable6 joined, notable6 joined, nativecallable6 joined, quotable6 joined, shareable6 joined, unicodable6 joined, bloatable6 joined, bisectable6 joined, sourceable6 joined 22:52 bloatable6__ joined 22:56 bloatable6 left 22:58 unicodable6 left 23:00 nativecallable6 left, quotable6 left 23:01 nativecallable6 joined, quotable6 joined, bloatable6__ left, bisectable6 left, sourceable6 left, unicodable6 joined 23:02 notable6 left 23:03 notable6 joined 23:04 bisectable6 joined, bloatable6 joined, sourceable6 joined, bloatable6__ joined, releasable6 left, shareable6 left 23:05 releasable6 joined, shareable6 joined 23:08 bloatable6 left 23:11 unicodable6 left, unicodable6 joined, notable6 left, nativecallable6 left, quotable6 left 23:12 bisectable6 left, sourceable6 left 23:13 bloatable6__ left, notable6 joined 23:14 releasable6 left, shareable6 left 23:15 quotable6 joined, nativecallable6 joined, releasable6 joined, shareable6 joined, shareable6__ joined, bisectable6 joined, sourceable6 joined 23:16 bloatable6 joined 23:19 shareable6 left 23:20 unicodable6 left 23:21 notable6 left 23:23 unicodable6 joined, quotable6 left, releasable6 left 23:24 bisectable6 left, sourceable6 left, notable6 joined 23:25 bloatable6 left, bisectable6 joined, sourceable6 joined 23:26 bloatable6 joined, nativecallable6 left 23:27 quotable6 joined, nativecallable6 joined, releasable6 joined 23:28 shareable6__ left 23:29 shareable6 joined, shareable6__ joined 23:33 unicodable6 left, unicodable6 joined, shareable6 left, notable6 left 23:34 bisectable6 left, sourceable6 left 23:35 bloatable6 left 23:36 quotable6 left, nativecallable6 left, releasable6 left 23:37 nativecallable6 joined, quotable6 joined, releasable6 joined, notable6 joined, bisectable6 joined, sourceable6 joined 23:38 bloatable6 joined, shareable6__ left, bloatable6__ joined 23:39 shareable6 joined, shareable6__ joined 23:41 unicodable6 left
Geth rakudo: ugexe++ created pull request #6681:
RakuAST: Evaluate package colonpairs of a hand built tree		 23:42
23:43 bloatable6 left, shareable6 left 23:44 unicodable6 joined 23:46 nativecallable6 left, quotable6 left, releasable6 left, notable6 left, bisectable6 left, sourceable6 left, releasable6 joined, notable6 joined, nativecallable6 joined, quotable6 joined 23:47 bisectable6 joined, sourceable6 joined, bloatable6__ left 23:48 shareable6__ left, bloatable6 joined 23:51 shareable6 joined, shareable6__ joined, unicodable6 left 23:54 releasable6 left, notable6 left, nativecallable6 left, quotable6 left, unicodable6 joined 23:55 shareable6 left 23:56 bisectable6 left, sourceable6 left 23:57 notable6 joined, quotable6 joined, nativecallable6 joined, bisectable6 joined, releasable6 joined, sourceable6 joined, bloatable6 left 23:58 bloatable6 joined, shareable6__ left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!