| ugexe | i guess i broke installed/precompiled zef again in one of my fixes | 03:20 | |
| can still test stuff, just using zef ala `RAKUDO_RAKUAST=1 path/to/bin/zef -I path/to/ install foo` | 03:22 | ||
| basically a cross-SC precomp issue related to the fact that methods get Callable[T] mixin behavior in rakuast, which causes precompiled zef to try and reference a type from a run time loaded module | 03:25 | ||
| i think that might require moarvm fixes though | |||
| need it to understand a STable is a parameterization and to rebuild via parametric_type.HOW.parameterize(params) at deserialization time | 03:29 | ||
|
07:39
sjm_ joined
07:53
sjm_ left
10:42
apogee_ntv left
10:43
apogee_ntv joined
|
|||
| Geth | rakudo/main: 434bb30b0c | (Elizabeth Mattijsen)++ | src/core.c/core_epilogue.rakumod Introduce RAKUDO_ASSUMING_DEBUG environment variable When set to 1, it will show the generated code deparsed. When set > 1, it will also show the generated AST Also make this code in .assuming a bit DRYer |
13:12 | |
| rakudo/main: cbc3852b21 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | 3 files Add native int/num method "bits" TIL I learned about the existence of nqp::objprimbits. This is now used to initialize $?BITS with, and to introduce the .bits method to native int and num types. Because these types do not directly exist as classes, the methods ... (8 more lines) |
13:16 | ||
| rakudo/main: ee30c34fff | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/Order.rakumod Add infix:<cmp>(Iterable:D, Iterable:D) In response to github.com/rakudo/rakudo/issues/6075 This will do a cmp on all the values that their iterators produce, and thus creating the same semantics for Seq cmp Seq or something with a Seq on either side, such as: dd (lazy (1,2,3)) cmp (1,2,3) which would produce `Less` instead of `Same` before this addition |
13:17 | ||
| rakudo/main: 9c322706ef | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/Duration.rakumod Make Duration based on an Int rather than Real Because we're dealing with nano-seconds really. Except there needs to be support for Inf and NaN as well (as there are spectest for these). So when an attempt is made to use Inf or NaN to make a Duration object with, instead return a glorified Rat that also provides a "tai" method to provide compatibility with Duration (and compatibility with spectests). Inspired by github.com/rakudo/rakudo/issues/6036 |
13:21 | ||
| [Coke] | (blin rakuast) - I assume we'd also need rakudos built with it set, which we don't have yet. | 13:29 | |
| (ah, catching up) -if it doesn't matter if the build was done with it, then the trick is making 'old' vs 'new' only change the env var. | 13:30 | ||
|
13:31
apogee_ntv left
|
|||
| Geth | rakudo/main: 46fc8602fb | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/IO/Path.rakumod Add :translate-nl named arg to IO::Path.slurp Defaults to True. So specify :!translate-nl if you don't want newline (\r\n -> \n) translation. Slurping a file without newline translation *can* make it up to 7% faster. As suggested in: github.com/rakudo/rakudo/issues/33...-569957370 |
13:31 | |
|
13:32
apogee_ntv joined
|
|||
| Geth | rakudo/main: d8609a46fb | (Will Coleda)++ (committed using GitHub Web editor) | 5 files Convert pre-compiled to precompiled The exception text is not specifically checked for in t or roast. Closes #6204 |
13:35 | |
| [Coke] | releasable6: next | 13:40 | |
| releasable6 | [Coke], Next release in ā33 days and ā5 hours. There are no known blockers. Changelog for this release was not started yet | ||
| [Coke], Details: gist.github.com/b10e9c9607a4fe948d...7f48a4d127 | |||
| Geth | rakudo/main: 9044b0d042 | (Nick Logan)++ (committed using GitHub Web editor) | 2 files Gate @_/%_ shadowing semantics on language revision Under 6.c/6.d both frontends keep doing what the legacy frontend has done since 2010: an @_/%_ reference inside a nested block (if/while/for body etc.) resolves outward to the enclosing routine's slurpy via lexical lookup. Lots of existing code is written against this. ... (18 more lines) |
13:44 | |
| lizmat | re speeding up compilation in RakuAST: I think if we could get rid of the Nodify function and instead be able to directly refer to the RakuAST:: classes in the Actions, that would already be a great help | 13:56 | |
| hmmm... looks like some recent commit broke App::Rak | 14:07 | ||
| ah, it broke Needle::Compile | 14:08 | ||
| Geth | rakudo/main: f62b5005bf | (Elizabeth Mattijsen)++ | src/Raku/Actions.nqp RakuAST: make handling of @_ and %_ DRYer |
14:21 | |
| lizmat | Needle::Compile broke after 2026.05 *phew* | 14:26 | |
|
14:30
librasteve_ joined
|
|||
| [Coke] | changelog draft up to date. | 14:34 | |
| lizmat | looks like 9044b0d042 broke it, investigating further | 14:45 | |
| linkable6 | (2026-05-25) github.com/rakudo/rakudo/commit/9044b0d042 Gate @_/%_ shadowing semantics on language revision | ||
| ugexe | That shadowing PR almost certainly has ecosystem fallout so a regular blin run would be useful already | 15:46 | |
| librasteve_ | rakudoweekly.blog/2026/05/25/2026-...lease-193/ | 15:56 | |
| ugexe | lizmat a guess is it might be related to the comment in github.com/lizmat/Needle-Compile/c...9396e2da58 | 16:56 | |
| i.e. it using a PointyBlock instead of whatever you originally wanted to use | |||
| lizmat | mayhaps | ||
| ugexe: gist.github.com/lizmat/ce9b35ab135...7eb718c202 fixes it | 17:15 | ||
| I'm sorta expecting the postfix if being the reason it fails | 17:16 | ||
| which would point to its scope being incorrect | |||
| ugexe | ah, yeah | ||
| lizmat | I mean, a postfix if should live in the same scope , no? | 17:17 | |
| ugexe | not really sure. unless im missing it, in the diff you showed the postfix stuff is inside a for loop which would have its own %_ | 17:21 | |
| lizmat | aah... yes... indeed | ||
| hmmm I wonder how many times this idiom lives in the ecosystem | 17:22 | ||
| ugexe | probably quite a bit, its mentioned in the commit message itself. but the idiom is bad so if we're ever going to get rid of it, it may as well be now | 17:23 | |
| the fix is trivial: use v6.d | |||
| lizmat | yeah, that feels about right | ||
| well, in that module's case, I would then need to add "use experimental :rakuast" :-) | 17:24 | ||
| ugexe | well luckily you're capable enough to not need the trivial fix :) | ||
| lizmat | interesting: "zef install Test::Coverage" segfaults for me | 17:25 | |
| hmmm.. .. any module, actually | 17:26 | ||
| lizmat nukes everything and re-installs | 17:27 | ||
| ugexe | yeah once you start using RAKUDO_RAKUAST=1 its pretty easy to get things in a bad state | 17:37 | |
| not so much start using the flag, but switching between frontends rather | |||
| lizmat | yeah, I recompiled various of the latest commits after each other when bisecting | 17:38 | |
| ugexe | i usually do make clean, rm -rf ~/.raku, rm -rf path/to/site path/to/core | 17:43 | |
| lizmat | rm -rf ~/.raku didn't fix it, I nuked install and rebuilt | 17:47 | |
| Fixed Needle::Compile uploaded | 18:00 | ||
| ugexe | gist.github.com/ugexe/0d3ad7d6c4fd...d34762f9c7 | 18:37 | |
| this is the golfed precomp issue that prevents installed zef from working | 18:38 | ||
| installed zef on a rakudo built with RAKUDO_RAKUAST=1 rather | |||
| i don't really know what to do from here though. i'm woefully unfamiliar with the moarvm stuff involved with that problem | 18:39 | ||
| lizmat | perhaps nine_ has some ideas | 18:41 | |
|
19:47
hurufu joined
21:18
apogee_ntv left,
apogee_ntv joined
21:42
hurufu left
21:59
apogee_ntv left
22:06
apogee_ntv joined
22:43
apogee_ntv left
22:49
apogee_ntv joined
22:54
apogee_ntv left
23:01
apogee_ntv joined
23:23
apogee_ntv left
23:24
apogee_ntv joined
|
|||