Geth rakudo: ugexe++ created pull request #6247:
Add an optimize phase to the RakuAST frontend
01:13
01:56 vrurg_ joined 01:58 vrurg left
Geth rakudo: ugexe++ created pull request #6248:
RakuAST: give onlystar bodies a source location
04:26
05:56 hurufu joined 06:13 [Coke] left, [Coke] joined
Geth nqp/main: e37642d997 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to get datagram leak fix, Zer0-Tolerance++
08:37
rakudo/main: 931393bdb9 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get datagram leak fix, Zer0-Tolerance++
09:15
rakudo/main: 6 commits pushed by (Nick Logan)++, (Elizabeth Mattijsen)++ 09:19
10:01 finanalyst joined 10:49 finanalyst left
Geth rakudo/main: 5f72bad53c | (Elizabeth Mattijsen)++ | src/Raku/ast/variable-declaration.rakumod
RakuAST: remove unused attribute

The string 'storage-name' didn't occur anywhere in the Rakudo code base except on this line. Was originally added with fdafab52bb in 2021 but appears to have never been used anywhere.
11:12
rakudo/lizmat-12: 2bce123617 | (Elizabeth Mattijsen)++ | 5 files
Another solution to fixing my %h is Set = 1,2,3

This effectively reverts e78c5a7e672786e5ae65 and adds logic to the codegenning to make sure "is Set", "is Bag" and "is Mix" are codegenned with `nqp::create(type)` rather then `type.new`. This ensures that it will always create a new instance. Then later the special versions of Set|Bag|Mix.STORE can return the sentinel ... (8 more lines)
12:00
rakudo: lizmat++ created pull request #6249:
Another solution to fixing my %h is Set = 1,2,3
12:01
12:21 librasteve_ left, ab5tract left, zostay_ joined, leont left 12:22 zostay left, zostay_ is now known as zostay, SmokeMachine left, tbrowder left 12:23 ab5tract joined, SmokeMachine joined 12:24 leont joined 12:25 tbrowder joined 12:36 librasteve_ joined
[Coke] + Bump NQP to get SQL output refactor fix, timo++ [c46a0f13] 13:36
^^ is this ultimately from moarvm? Don't see a changelog entry in moarvm for it.
ah, no, it's directly from nqp found it
all commits so far in the changelog. 13:41
as usually, mostly just a 1:1 from commits, with some merge commits tacked on to one of its included commits.
github.com/rakudo/rakudo/wiki/ChangeLog-Draft 13:42
ugexe just to be clear my current intention is to not work on optimizer stuff, but i thought putting some baby optimizations in place would open the door for other people to add more 14:10
my focus will still generally be on fixing rakuast issues exposed by the ecosystem 14:11
[Coke] ugexe++ 14:32
Geth rakudo/lizmat-13: d135ec9e81 | (Elizabeth Mattijsen)++ | src/Raku/ast/variable-declaration.rakumod
RakuAST: streamline sigil -> role mapping

Make the lookup hash a constant, and only look up the sigil once.
Sadly this is too early to make the hash contain RakuAST::Name objects, so we just keep the name and do the instantiation of RakuAST::Name when needed.
17:04
rakudo: lizmat++ created pull request #6250:
RakuAST: streamline sigil -> role mapping
rakudo/lizmat-14: 66f03215af | (Elizabeth Mattijsen)++ | 7 files
RakuAST: make some lookup hashes set at compile time

So they don't need to be re-created and initialized on each call.
17:25
rakudo: lizmat++ created pull request #6251:
RakuAST: make some lookup hashes set at compile time
17:27
lizmat argh... that also included lizmat-13
ah well
it was just for the RakuAST CI tests really
Geth rakudo/main: 336e76501a | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
RakuAST: give onlystar bodies a source location

A proto's onlystar body compiles to a frame whose only code is the dispatch instruction, and that frame carried no location annotation. The OnlyStar node is created directly in the routine actions rather than through an action method of its own, so it never received an origin, and its QAST carried no node to annotate from. Code.file and ... (16 more lines)
17:34
rakudo/main: 0918c0552e | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | 5 files
Another solution to fixing my %h is Set = 1,2,3

This effectively reverts e78c5a7e672786e5ae65 and adds logic to the codegenning to make sure "is Set", "is Bag" and "is Mix" are codegenned with `nqp::create(type)` rather then `type.new`. This ensures that it will always create a new instance. Then later the special versions of Set|Bag|Mix.STORE can return the sentinel ... (8 more lines)
17:39
roast: 26fc8d9736 | (Elizabeth Mattijsen)++ | 3 files
Add tests for "is Set|Bag|Mix" keeping sentinels clean

See github.com/rakudo/rakudo/issues/6246
17:49
rakudo/main: 69b2592291 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/Array.rakumod
Fix Array.shift on holes in arrays

Unlike the needed Array.pop fix in 415fa4b069 to prevent a segfault, Array.shift was already guarded against returning an nqp::null.
However, it was set to return Nil, rather than the default value of the array. This commit changes that to be more in line with ... (10 more lines)
17:53
roast: b87c37fc1c | (Elizabeth Mattijsen)++ | S32-array/shift.t
Fix some .shift tests

1. github.com/rakudo/rakudo/pull/6242 changed the behaviour
   of .shift on a hole to returning the default value of the array
   rather then Nil (also what one would get if one would do @a[0]
   on a hole)
2. Activate test of shifting off an infinite array: that apparently
   works nowadays :-)
18:03
rakudo/main: b0c6b1b9f8 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | 8 files
RakuAST: make some lookup hashes set at compile time

So they don't need to be re-created and initialized on each call.
18:11
18:23 hurufu1 joined, hurufu left, hurufu1 is now known as hurufu
Geth nqp/main: 04a224de3e | (Nick Logan)++ (committed using GitHub Web editor) | 4 files
HLL::Compiler.compile: balance backend per-compile state on early exit

HLL::Backend.start (MoarVM) pushes a frame list onto
  %COMPILING<moar><frames> at every compile's 'start' stage. The matching
pop runs at the 'mast' stage in QASTCompilerMAST.to_mast. A compile that stops before 'mast' (e.g. with `:target('ast')`), or one whose stage call throws, leaves the push dangling. The next nested compile's ... (16 more lines)
18:58
rakudo/main: 25200ba632 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get balanced backend state on early exit

  ugexe++
19:08
lizmat m: say nqp::const::CCLASS_ANY # apparently, nqp::const::xxx does *not* need "use nqp" 20:09
camelia 65535
lizmat TIL
20:10 finanalyst joined
[Coke] that seems odd 20:17
Geth rakudo/lizmat-15: 6ef106b169 | (Elizabeth Mattijsen)++ | src/Raku/Actions.nqp
RakuAST: some streamlining in actions

1. prevent repeated named capture lookups 2. streamline is-yada to not use a regular expression, but more
   native logic to check if a package is a stub
20:57
rakudo: lizmat++ created pull request #6252:
RakuAST: some streamlining in actions
21:52 hurufu left 22:03 finanalyst left