🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
03:01 MasterDuke joined
MasterDuke huh. well that whole exercise was somewhat unnecessary. turns out the jvm build is fine at HEAD of nqp/rakudo. however, it may require openjdk 11 (instead of openjdk 21 as was my system default) and/or `make -j1` (instead of `make -j12`, like i usually use) 03:03
hm. but my fastutil branch doesn't seem to build even with `make -j1`...ugh 04:21
06:53 Util left 07:00 Util joined 08:21 timo left 08:58 timo joined 09:19 sena_kun joined
Geth nqp/main: 4262be61ea | (Elizabeth Mattijsen)++ | 2 files
Add SIG_ELEM_ALL_NAMES_OK + SIG_ELEM_SLURPY_ARITY

And fix SIG_ELEM_IS_NOT_POSITIONAL on the JVM backend
09:34
rakudo/main: 2f67432810 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bum NQP to get new nqp::const values

  - SIG_ELEM_ALL_NAMES_OK
  - SIG_ELEM_SLURPY_ARITY
09:50
11:01 timo left 11:15 timo joined 11:27 timo left, timo joined
Geth rakudo/main: 8f9ad9a3ef | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Streamline Routine quite a bit

Including the sort_dispatchees logic. This causes "make test" to drop from 202 CPU seconds -> 195 CPU seconds, and lose a few seconds wallclock on "make spectest", as well as installing core modules < 8 seconds wallclock.
... (5 more lines)
11:52
lizmat actual dispatch logic next
[Coke] MasterDuke: if you can add whatever is state of the art needed for build in the jvm readme, it'll help the occasional dabbler like myself. 12:56
Geth ¦ rakudo: coke assigned to lizmat Issue discrepancy in --doc output between standard/RAKUAST github.com/rakudo/rakudo/issues/5536 13:12
[Coke] understands if the answer is "we'll get the bug fix when RakuAST lands" 13:13
lizmat I think the answer is: the old renderer gets it wrong 13:14
:-)
Geth rakudo/main: 8b00ff5b70 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Remove "add_to_cache" functionality from Moar backend

Instead of just blocking out the contents of the method
14:18
15:26 MasterDuke left
Geth rakudo/main: 0b65e78b44 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Add Routine.dispatch_order accessor

Which will sort the dispatchees on demand of not sorted already
17:15
rakudo/main: 3a66dc104d | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Streamlining Routine.find_best_dispatchee (Part 1)

This is mostly about de-indenting the code, and localizing some variables to the scope that their needed.
17:59
rakudo/main: c9ecbda2b7 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Streamlining Routine.find_best_dispatchee (Part 2)

We only need one flag to indicate mismatches
18:13
lizmat going to do the streamlining of find_best_dispatchee very carefully and in small steps 18:18
as this is really deep at the heart of Raku, and if something gets borked, (and it undoubtedly will), we can have a better bisect
jdv is this mostly part of "new disp" or does it predate that? 18:20
lizmat this goes all the way back to nom 18:23
with some additions for unsigned ints since then
Geth rakudo/main: 48509e00b8 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Streamlining Routine.find_best_dispatchee (Part 3)

Flip the meaning of the mismatch flag, so we don't have to negate when checking. For changing the value, it doesn't matter whether 1 or 0 is set.
18:25
rakudo/main: 70b8b229fa | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Streamlining Routine.find_best_dispatchee (Part 4)

Some more object vs native checking
19:08
19:50 MasterDuke joined
MasterDuke [Coke]: i don't think anything terribly special is needed, though i guess i could try a couple java versions other than 11 and 21 and mention which ones work 19:52
ah, the nqp readme says 9 is the minimum, so i guess i just need to try 17 and 21 (since we know 11 works) 19:57
[Coke] +1 21:06
I haven't had luck with os x java (Which is I think is the oracle version) 21:07
don't think I've built a java build since I got the mac.
lizmat fwiw, I gave up on building on the JVM backend on MacOS 22:56
MasterDuke the mac jvm ci jobs are passing, so it must be possible to build somehow 23:21
looks like it's using java 11 23:25
23:37 sena_kun left