|
02:02
guifa left
03:16
apogee_ntv left
03:18
apogee_ntv joined
|
|||
| Geth | rakudo: ugexe++ created pull request #6163: Make CORE/SETTING symbol tests pass on RakuAST built rakudo |
04:57 | |
|
06:06
bloatable6__ left,
Voldenet left,
lizmat left,
coleman left,
evalable6 left,
gfldex left,
disbot2 left,
kurahaupo left,
ShimmerFairy left,
summerisle left,
[Coke] left,
rba left,
SmokeMachine left,
JRaspass left
06:10
kurahaupo joined,
bloatable6__ joined,
Voldenet joined,
ShimmerFairy joined,
[Coke] joined,
lizmat joined,
rba joined,
coleman joined,
evalable6 joined,
gfldex joined,
SmokeMachine joined,
disbot2 joined,
summerisle joined,
JRaspass joined
10:20
librasteve_ joined
10:27
guifa joined
10:50
guifa left
10:52
guifa joined
10:56
guifa left
10:59
guifa joined
|
|||
| Geth | rakudo/main: 6cb6ac321d | (Elizabeth Mattijsen)++ | 2 files Re-instate the RakuAST based implementation of .assuming This had to reverted in March 2025 because of problems with precompilation of synthetic RakuAST code. The problem was fixed by ugexe++ in: github.com/MasterDuke17/rakudo/com...6301b348a8 For more information, see: dev.to/lizmat/quicker-to-assume-2cpj |
11:01 | |
| lizmat | librasteve_ ^^ | 11:02 | |
| [Coke]: ^^ that'd be worth an extra blin run | 11:03 | ||
| as it was a blin run that initiated the revert :- | |||
|
11:04
guifa left
|
|||
| lizmat | several tests in roast now passing, but won't untodo them until it's clear that this revert survives a blin run :-) | 11:12 | |
|
11:37
Voldenet left,
Voldenet_ joined,
Voldenet_ is now known as Voldenet
11:38
[Coke] left
|
|||
| lizmat | Geth | 11:39 | |
|
12:05
[Coke] joined
|
|||
| [Coke] | lizmat: sure. | 12:33 | |
| Geth | rakudo/main: 81b3d1721b | (Elizabeth Mattijsen)++ | src/core.e/Format.rakumod sprintf in 6.e is now RakuAST based The temporary *undocumented* "zprintf" is gone now. |
12:39 | |
| [Coke] spends some time cleaning up the disk on the blin machine (so much leftover stuff after a run, disk nearly full after the last run for some reason) | 12:53 | ||
| started blin | 13:05 | ||
| lizmat | offline for an upgrade | 13:16 | |
|
13:16
lizmat left
|
|||
| [Coke] ponders a "when was the last version of rakudo that a module worked on" script. | 13:30 | ||
| If it's > x releases ago, skiplist it for blin. They can ask to be re-added to the list when they pass with a recent one | |||
|
13:32
shareable6 left
13:34
shareable6 joined
13:37
lizmat joined
|
|||
| Geth | rakudo/main: 8bf0321569 | (Nick Logan)++ | src/Raku/ast/variable-declaration.rakumod RakuAST: declare $¢ with a Scalar container, matching legacy RakuAST::VarDeclaration::Implicit::Cursor was emitting QAST::Var with :decl('var') — a bare lexical slot, no container. After CORE.d.setting was compiled, SETTING:: at v6.d had a slot named '\$¢' but nqp::getlexrel returned null for it, because no container had been installed at the slot. ... (10 more lines) |
13:41 | |
| rakudo/main: c6c2cb1f80 | (Nick Logan)++ | 7 files Account for CORE/SETTING differences when built with RakuAST When CORE is built by the RakuAST frontend, three symbols differ from the legacy frontend at SETTING:: and CORE:: scope: - !RAKUAST_MARKER is a RakuAST-only marker; legacy never installed it - $?FILE is a lexical constant in the RakuAST CORE setting pad; legacy ... (13 more lines) |
|||
| rakudo/main: e20e8cc08f | (Nick Logan)++ (committed using GitHub Web editor) | 8 files Merge pull request #6163 from ugexe/ugexe/rakuast-corekeys-settingkeys-tests Make CORE/SETTING symbol tests pass on RakuAST built rakudo |
|||
| rakudo: ugexe++ created pull request #6164: Add a rakuast variant to the Azure CI build/test matrix |
|||
|
13:47
lizmat left
13:48
lizmat joined
|
|||
| Geth | roast: af0efdd92d | (Elizabeth Mattijsen)++ | 2 files Mark test as failing on MacOS Tahoe They were already marked as failing on Sonoma and Sequoia |
14:27 | |
| rakudo/main: 6ae9ef82dd | (Elizabeth Mattijsen)++ | 4 files Remove zprintf from core keys tests The sub was already removed from core in 81b3d1721b |
14:31 | ||
| rakudo/main: 1e1348145e | (Elizabeth Mattijsen)++ | src/core.e/Formatter.rakumod Fix issue with sprintf("%e") formatting in 6.e The new RakuAST based implementation of sprintf had an issue with exact multiples of 10, and didn't show integer values correctly. This has now been fixed. The was one of the listed TODOs from 2023. The other todo I decided to ignore, as sprintf("%s",Mu) is legal, so why isn't sprintf(%f,Mu) either? github.com/Raku/old-issue-tracker/issues/4537 |
15:49 | ||
| lizmat | m: dd Date.today.fmt | 16:15 | |
| evalable6 | (exit code 1) Cannot resolve caller fmt(Date:D); Routine does not have any candidates. Is only the proto defined? in block <unit> at /tmp/vqBqrC_nS6 line 1 |
||
| lizmat | meh | ||
| m: dd (Date.today,).fmt | |||
| evalable6 | "2026-05-06" | ||
| lizmat | that feels inconsistent | ||
| m: dd (Date.today,).fmt("%s") | 16:35 | ||
| evalable6 | "2026-05-06" | ||
| Geth | rakudo/main: 505c6b4f00 | (Nick Logan)++ | azure-pipelines.yml Add a rakuast variant to the Azure CI build/test matrix Builds and tests rakudo with RAKUDO_RAKUAST=1 on Linux/MoarVM so the new frontend keeps getting exercised on every push, preventing regressions until it becomes the default. |
||
| rakudo/main: 1e6da94c03 | (Nick Logan)++ (committed using GitHub Web editor) | azure-pipelines.yml Merge pull request #6164 from ugexe/rakuast-ci-variant Add a rakuast variant to the Azure CI build/test matrix |
|||
| lizmat | ugexe++ | ||
|
17:03
ugexe joined
|
|||
| Geth | roast: b931ef94f2 | (Elizabeth Mattijsen)++ | 2 files Activate 6.e sprintf "%c" testing It is without errors atm, so no reason not to include it |
18:01 | |
|
20:36
guifa joined
20:49
lizmat left,
lizmat_ joined
20:51
lizmat_ left,
lizmat joined
21:13
guifa left
|
|||
| [Coke] | (blin) Hurm. Easy change for the skiplist is not never run it, but "don't bother testing the old version if it's the skiplist" that's like a 30-40% speedup right there. | 21:36 | |
| (if it *failed* and is in the skiplist) | |||
| (*#@&4. "let's git clean this to get rid of all the crap". Removing TODO.md ... NOTES.md ... | 21:54 | ||
| whoops | |||
| Geth | Blin/coke/rm-pakku: 30bbd2a3a1 | (Will Coleda)++ | 3 files Remove Pakku Historically, having access to both was great, but zef is the de facto winning package manager. |
21:56 | |
| Blin: coke++ created pull request #46: Remove Pakku |
|||
| rakudo: ugexe++ created pull request #6165: Fix @_/%_ resolution inside nested blocks under RAKUDO_RAKUAST=1 |
21:57 | ||
| Blin/coke/cleanup: 11 commits pushed by (Will Coleda)++ review: github.com/Raku/Blin/compare/eee2f...0e523a1b52 |
22:22 | ||
| rakudo/main: d8957e5430 | (Nick Logan)++ | src/core.c/core_epilogue.rakumod RakuAST: don't rely on magical %_ inside nested blocks of ParameterAST ParameterAST in core_epilogue.rakumod accessed %_<type>, %_<name> and %_<default> from inside if/else block bodies. Under 6.e semantics each block is a "real block with its own signature" and a reference to %_ inside one auto-declares a fresh empty *%_ on that block, shadowing the outer one. Rename the slurpy to a non-magical *%opts so the function ... (6 more lines) |
22:36 | ||
| rakudo/main: b267180383 | (Nick Logan)++ | src/Raku/Actions.nqp RakuAST: align @_/%_ resolution with the legacy frontend's de-facto behavior The slurpy-placeholder action installed a fresh *@_/*%_ on every block that referenced @_/%_, including non-routine blocks like if/while/for bodies. So inside `sub f(*%_) { if 1 { %_<k> } }` the inner reference resolved to a fresh empty placeholder on the if-body instead of f's %_, breaking core_epilogue's .assuming (and the rest of the ... (8 more lines) |
|||
| rakudo/main: 9a66df209e | (Nick Logan)++ (committed using GitHub Web editor) | 2 files Merge pull request #6165 from ugexe/ugexe/fix-rakuast-placeholder-shadowing Fix @_/%_ resolution inside nested blocks under RAKUDO_RAKUAST=1 |
|||
| rakudo: ugexe++ created pull request #6166: 6.e: Gate @_/%_ shadowing semantics on language revision |
22:39 | ||
| [Coke] | whoops, forgot to run this blin inside tmux, thankfully my laptop battery didn't die, | 23:12 | |
| Geth | Blin/coke/cleanup: 2e5f75bd97 | (Will Coleda)++ | 3 files Move list of modules to a JSON file. |
23:14 | |
|
23:39
librasteve_ left
|
|||