Geth rakudo: ugexe++ created pull request #6829:
RakuAST: worry about any primed expression in sink context
00:01
rakudo/main: 40c2c633b1 | (Nick Logan)++ | 2 files
RakuAST: make any bare item of an enum list a key

Previously a bare item in the list of an enum declaration only became a key when it was a Str. Any other item, such as the Int of
  `enum F (5.succ)`, was dropped, and an enum made only of such items had
no base type, so building it died with "Cannot find method 'mro'". This takes any item that is not a Pair as a key, which is made a Str and given the next value, as the legacy frontend does.
01:58
rakudo/main: 86847b4c15 | (Nick Logan)++ | 2 files
Make the Map an enum declaration evaluates to match the keys of the enum

Previously the Map an enum declaration evaluates to took each key as it was written and flattened the items of the enum list. So an enum with a key that is not a Str, such as `enum F (6)` or `enum F (6 => 1)`, died with "This type cannot unbox to a native string" once its value was used, and `enum F (<a b>, <c d>)` gave a key for each word where the enum has one for each list. This takes the items of the list and makes a key that is not a Str a Str, the same way the enum does with its own keys.
rakudo/main: b355a7c1f5 | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
Merge pull request #6823 from ugexe/ugexe/rakuast-enum-non-str-keys

RakuAST: make any bare item of an enum list a key
rakudo/main: 28322ca46e | (Nick Logan)++ (committed using GitHub Web editor) | 2 files
RakuAST: evaluate the shape of an inline array as an expression (#6824)

Previously the shape of a `HAS` attribute that is a CArray was evaluated at BEGIN time twice, as a statement list, which can only be interpreted. So a shape that has to be compiled, such as `(4, 1)[0]`, `do { 4 }` or one with a statement modifier, died with "Cannot unbox a type object
  (NQPMu) to an int", which also hid the error of a shape that dies. This
evaluates the shape once, as a parenthesized expression, and only sizes the array when that evaluation succeeds.
rakudo/main: 94de42d9c4 | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
RakuAST: keep the condition of an until loop as written (#6825)

Previously an until loop, a repeat until loop or an until statement modifier that compiles to a `from-loop` call, as one whose value is wanted or one with an UNDO phaser does, had its condition replaced at BEGIN time with the condition wrapped in a `.not` call, which is how the negation reached the `from-loop`. Anything looking at the tree afterwards ... (5 more lines)
rakudo/main: e80e60d827 | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
RakuAST: give xx the parameters of a WhateverCode operand it primes (#6827)

Previously `xx` primed a WhateverCode operand built from an infix, but left the parameters of that WhateverCode behind. `"a" xx * - 1` and
  `* - 1 xx 2` gave a WhateverCode that took no arguments and died when
called, and since 31d48fde2d primes `1 Rxx *`, so did `1 Rxx * xx 2`. The search for the arguments of the new WhateverCode stopped at an ... (15 more lines)
rakudo/main: 4736c94899 | (Nick Logan)++ | 2 files
RakuAST: give a signature literal the span of its delimiters

Previously the origin of a signature literal such as `:($x, $y)` or the
  `:(Int)` of a `&c:(Int)` parameter covered only its parameters, as the
node is made where the signature is parsed, inside `:(` and `)`, and the block of its parameters had no origin at all. This widens both over the whole literal.
rakudo/main: d41bd7f387 | (Nick Logan)++ | 7 files
RakuAST: claim the leading doc of a subset or an enum before its expression

Previously a subset or an enum became the declarand only once its where clause or term was parsed, so a block or other declarand in that expression took its leading doc first. Only a where clause that is itself a block was handled, by taking its doc back. `#| doc` before
  `subset G of Int where ({ $_ > 0 })`,
... (19 more lines)
rakudo/main: c1844e3e22 | (Nick Logan)++ (committed using GitHub Web editor) | 8 files
Merge pull request #6828 from ugexe/ugexe/rakuast-subset-where-nested-doc

RakuAST: claim the leading doc of a subset or an enum before its expression
rakudo/main: 0cdbad0709 | (Nick Logan)++ (committed using GitHub Web editor) | 2 files
RakuAST: worry about any primed expression in sink context (#6829)

Previously a primed expression in sink context drew a useless use worry only when its operator was pure, as with `* + 1;`. Since 31d48fde2d a meta-operator primes on every side, so `@a Z= *;` and `$x [R//]= *;` build a WhateverCode and discard it where they used to assign, with no worry. `$x += *;`, `*.succ;`, `* ~~ Int;` and `++*;` drew none either, ... (11 more lines)
05:23 hurufu joined, kjp left 05:27 kjp joined 06:04 [Coke] left 06:07 [Coke] joined 06:29 finanalyst joined 08:43 hurufu left 09:05 lizmat joined
lizmat . 09:05
13:20 hurufu joined 13:48 hurufu left 13:49 hurufu joined
japhb Got a failure from an overnight build on Jupyter::Chatbook -- Looks like a slang issue 13:49
13:53 hurufu left 13:54 hurufu joined
Geth rakudo: ugexe++ created pull request #6830:
Fix several edge cases of || in 6.e subscripts
14:05
14:19 hurufu left 14:46 hurufu joined 15:31 hurufu left 15:36 hurufu joined 15:56 hurufu left 15:57 hurufu joined
[Coke] 74b7282e15d984563c3c 16:14
linkable6 (2026-10-07) github.com/rakudo/rakudo/commit/74b7282e15 RakuAST: declare a package's body with the block it sits in
[Coke] github.com/coke/raku-blin-release-results - 2 errors on ^^ commit. 17:35
tested up through 6a3feeac14
linkable6 (2026-10-08) github.com/rakudo/rakudo/commit/6a3feeac14 RakuAST: apply a condition modifier inside the loop modifier's body (#6821)
18:01 hurufu left 18:07 hurufu joined 20:05 rnddim joined, rnddim left 20:06 ShimmerFairy left, Pixi left, rnddim joined 20:08 rnddim left, rnddim joined 21:34 hurufu left