Geth rakudo: ugexe++ created pull request #6610:
RakuAST: make native arguments behave like container arguments
01:39
rakudo/main: 6 commits pushed by (Nick Logan)++ 02:22
02:46 ilogger2 left 02:59 ilogger2 joined
[Coke] releasable6: next 03:47
releasable6 [Coke], Next release in ≈27 days and ≈15 hours. There are no known blockers. 49 out of 86 commits logged
[Coke], Details: gist.github.com/c203370757d5e3f3c8...4c05e4de80
[Coke] well, restarting the bots seems to have helped slightly
Geth rakudo: ugexe++ created pull request #6611:
RakuAST: look up routines the syntax calls by a static lookup
04:14
rakudo/main: 67c81bd6d6 | (Nick Logan)++ (committed using GitHub Web editor) | 8 files
RakuAST: look up routines the syntax calls by a static lookup (#6611)

Previously the static callee mark covered calls by name and infix, prefix, and postfix operators. A subscript, an array or hash composer, a term that names a routine, a list operator, a capture variable, a smartmatch, and the operator beneath an assignment, reverse, or sequence meta-op still compiled as a call by name. The ... (13 more lines)
04:56
rakudo: ugexe++ created pull request #6612:
RakuAST: pass the paired literal of a negated operator natively
05:14
rakudo: ugexe++ created pull request #6613:
RakuAST: reduce a reversed operator with reversed associativity
05:36
09:17 finanalyst joined 10:33 finanalyst left
Geth rakudo/main: 20ab5ae8fa | (Elizabeth Mattijsen)++ | src/core.c/ThreadPoolScheduler.rakumod
Remove one %*ENV lookup from startup

These cases only look for something in there, so don't need val() processing. This will allow %*ENV to be set up later during startup.
12:54
rakudo/main: edeae9638c | (Elizabeth Mattijsen)++ | 2 files
Move setting $*SCHEDULER to much later in the setting

Because the setting itself doesn't need it, and this opens up some other optimization possibilities.
13:02
rakudo/main: 53496f247b | (Nick Logan)++ (committed using GitHub Web editor) | 3 files
RakuAST: reduce a reversed operator with reversed associativity (#6613)

Reversing an operator reverses its associativity, so a reduce of it folds the other way. Previously a reduce of any reversed operator folded left, which made [R-] 1, 2, 3 produce 2 instead of 0.
This takes the reducer from the reversed operator properties the node ... (8 more lines)
14:59
15:07 djinni` left 15:29 djinni` joined 18:20 kurahaupo left 18:32 ilogger2 left, ilogger2 joined 19:12 samebchase7 joined 19:13 Geth__ joined 19:15 sjn_ joined 19:18 samebchase left, jjatria left, Geth left, sjn left, samebchase7 is now known as samebchase 19:19 jjatria joined 19:30 kurahaupo joined 19:39 MasterDuke joined 19:40 MasterDuke left 19:41 MasterDuke joined 19:44 MasterDuke left, MasterDuke joined
MasterDuke ugexe: i was just thinking that a potentially good project for your ai assistants is going through the last two years of irc logs and looking for the things nine brought up as major pain points. i feel like there were a whole bunch of things that he classified as the second 90%. i.e., the really difficult rakuast work to get to 100% roast, but for 19:50
really minor gain. unfortunately i don't have any concrete examples in mind, but this sort of brute force log parsing/sentiment analysis seems like a good fit for ai
then those are potential candidates for removal (in 6.f, 6.e?, tbd) to provide a significantly simpler language implementation at minimal loss of language capabilities 19:53
20:15 finanalyst joined 20:22 MasterDuke left
ugexe by 100% roast coverage, do you mean things marked todo that aren't passing? we otherwise already pass `make spectest` 20:50
21:48 finanalyst left 22:45 azert joined 22:49 azert left, azert joined 22:50 azert left
Geth__ rakudo: ugexe++ created pull request #6614:
Sort, slip and split without work the answer does not need
22:56