|
🦋 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. |
|||
|
00:00
sena_kun left
00:05
kjp left
00:08
kjp joined
01:19
MasterDuke joined
|
|||
| MasterDuke | . | 01:20 | |
| tellable6 | 2024-03-01T20:31:16Z #raku-dev <[Coke]> masterduke - Tried building HEAD and 2024.01, both die with guarantee(requested_word_size <= chunklevel::MAX_CHUNK_WORD_SIZE) failed: Requested size too large (620714) - max allowed size per allocation is 524288. | ||
| 2024-03-01T20:31:39Z #raku-dev <[Coke]> Masterduke this on OSX w/ javac 20.0.2 | |||
| MasterDuke | [Coke]: IIRC, java 20 has a known problem that was introduced in 17 and wasn't fixed until 21 | 01:21 | |
|
08:40
sena_kun joined
09:34
sena_kun left
|
|||
| Geth | rakudo/main: 97a0eeab69 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp Streamline bind_one_param in the bootstrap - abstract concreteness / typecheck error handling into separate subs - reorder some conditionals for more efficient throughput - don't create error message for junction returns - add some more comments In preparation for more thorough streamlining |
13:37 | |
| rakudo/main: 18989a661b | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp Make order of positionals bind/bind_one_param match Mostly for readability / orderedness :-) Also move helper sub to the helper sub area for easier browsing |
13:54 | ||
|
15:02
sena_kun joined
|
|||
| MasterDuke | i haven't been doing a proper bisect, just stepping by nqp bumps, but so far rakudo has built up to 5c9156e268b103997f451f0ed2d0f4ebcd9e6698 (i had to cherry-pick my unbreak the jvm build nqp commit for the last couple bumps, but that was expected) | 16:04 | |
| linkable6 | (2024-02-12) github.com/rakudo/rakudo/commit/5c9156e268 Bump NQP to get latest MoarVM tweaks | ||
| MasterDuke | heh. and now i've gotten myself into a situation where the moarvm rakudo build failed, but the jvm build seems to be soldiering on... | 16:17 | |
| lizmat | MasterDuke: 6583e61c5b265257 broke the build | 18:03 | |
| and was reverted | |||
| could that be it? | |||
| which leaves 94ee052770e28adc03e21 , which was a MoarVM bump | 18:04 | ||
| linkable6 | (2024-02-12) github.com/Raku/nqp/commit/94ee052770 Bump MoarVM for the latest tweaks/optimizations | ||
| lizmat | which basically means 6a9c523dd5145d4c2d | 18:06 | |
| linkable6 | (2024-02-11) github.com/MoarVM/MoarVM/commit/6a9c523dd5 Slightly optimize creating backtraces | ||
| lizmat | or a1a18e9842464fdc17a | ||
| linkable6 | (2024-02-12) github.com/MoarVM/MoarVM/commit/a1a18e9842 Fix uninitialized is_unsigned flag in P6num | ||
| MasterDuke | the moarvm thing is probably just an artifact of something not getting quite cleaned up completely. i am playing fast and loose with checking out various commits, cherry-picking others, etc | 18:11 | |
| lizmat | .oO( playing fast and loose: the SpaceX way :-) |
18:16 | |
| MasterDuke | as long as there aren't any rapid unscheduled disassemblies in my house... | 18:18 | |
| jdv | used to be the fb thing until zuck realized he was young and dumb | 18:22 | |
| MasterDuke | Voldenet: i was surprised by your statement that memory use on my fastutil branch was higher. how did you measure? | 19:10 | |
| Voldenet | I measured the memory consumption during build using mem_usage_ui | 19:13 | |
| MasterDuke | huh. not at all what i was expecting | 19:15 | |
| Voldenet | I think the difference is too substantial to be coincidence | 19:18 | |
| plots of compilations imgur.com/a/uYLBepJ | |||
| iirc I didn't get full profile of raku-ast-compiler, but the difference is clearly seen | 19:19 | ||
| lizmat | am I reading this correctly that the RakuAST compiler takes way less memory ? | 19:21 | |
| MasterDuke | hm. well i think there were one or two commits that didn't use fastutil but i thought would help, i'll try to measure them independently | ||
| Voldenet | Yes | ||
| but it takes a lot more time | |||
| (both profiles are not full, I didn't plan to use them for anything) | 19:22 | ||
| v6c consumes similar amount of memory | |||
| I'm speculating that it takes less memory in general, but some frequent list/dict gets a lot bigger than it needs to | 19:31 | ||
| I might re-run that later while doing more accurate profiling, for now I simply tried to build it all | 19:34 | ||
| MasterDuke | could you try with just the change in github.com/MasterDuke17/nqp/commit...8b81df0bfd (but done to stock nqp, not my branch)? | 19:38 | |
| probably afk for a while & | 19:39 | ||
| lizmat | greppable6: .AUTOTHREAD | 19:45 | |
| greppable6 | lizmat, Found nothing! | ||
| lizmat | greppable6: AUTOTHREAD | 19:47 | |
| greppable6 | lizmat, Found nothing! | ||
| [Coke] | MasterDuke: thanks | 20:07 | |
|
20:53
japhb left
21:02
japhb joined
22:40
sena_kun left
|
|||
| Geth | rakudo/main: fb7e44fcd5 | (Elizabeth Mattijsen)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp Streamline bind_sig, is_bindable, bind_cap_to_sig |
23:37 | |