🦋 Welcome to the MAIN() IRC channel of the Raku Programming Language (raku.org). Log available at irclogs.raku.org/raku/live.html . If you're a beginner, you can also check out the #raku-beginner channel!
Set by lizmat on 6 September 2022.
00:38 lichtkind__ left 00:40 constxqt left 00:45 constxqt joined 00:53 constxqt left 01:19 constxqt joined 01:25 constxqt left 01:39 constxqt joined 01:48 constxqt left 02:12 constxqt joined 02:18 constxqt left
elcaro The linux download on rakudo.org/downloads is halting at 16.2M (~82%) 02:23
02:24 constxqt joined 02:29 constxqt left 02:37 constxqt joined
coleman can you paste the exact link you're using elcaro 02:47
02:47 constxqt left
coleman i can't reproduce 02:50
02:54 constxqt joined 02:55 hulk joined 02:57 kylese left 03:08 constxqt left 03:09 constxqt joined 03:13 Voldenet_ joined, constxqt left 03:14 Voldenet left, Voldenet_ is now known as Voldenet 03:24 constxqt joined 03:29 constxqt left 03:30 hulk left 03:34 kylese joined 03:41 constxqt joined 03:46 constxqt left 03:57 constxqt joined 04:01 constxqt left 04:02 constxqt joined 04:12 constxqt left 04:13 Voldenet_ joined 04:14 Voldenet left, Voldenet_ is now known as Voldenet 04:15 constxqt joined 04:26 constxqt left 04:53 constxqt joined 04:59 constxqt left 05:11 constxqt joined 05:15 constxqt left 05:29 constxqt joined 05:40 constxqt left, constxqt joined
elcaro rakudo.org/dl/rakudo/rakudo-moar-2...gcc.tar.gz 05:45
wget from multiple linux boxes gets "Connection closed" after 16M. Also tried downloading via Firefox on Windows and it also fails 05:48
05:48 avuserow joined 05:50 constxqt left
avuserow elcaro: works for me on two different ISPs. Though if I download via IPv6 it is slow (~500KB/s vs ~60MB/s on IPv4), so maybe try `wget -4 ...`? 05:52
06:19 constxqt joined 06:24 constxqt left 06:36 constxqt joined 06:41 constxqt left 06:58 constxqt joined 07:03 constxqt left 07:11 dbonnafo left, dbonnafo joined 07:15 constxqt joined 07:19 dbonnafo left, dbonnafo joined 07:25 constxqt left 07:42 constxqt joined 07:49 constxqt left 08:06 constxqt joined 08:16 constxqt left
Geth advent/main: 7593e0e9e2 | (Nigel Hamilton)++ (committed using GitHub Web editor) | raku-advent-2023/authors.md
Update authors.md
08:38 constxqt joined 08:43 dbonnafo left 08:47 constxqt left 08:49 dbonnafo joined 08:51 lichtkind__ joined 08:52 dbonnafo left 08:54 jpn joined, dbonnafo joined 09:00 constxqt joined 09:05 constxqt left 09:08 jpn left 09:29 constxqt joined 09:34 constxqt left 09:44 sena_kun joined 09:58 japhb left 10:02 constxqt joined 10:07 constxqt left, jpn joined 10:16 jpn left 10:20 constxqt joined 10:25 constxqt left 10:37 constxqt joined 10:42 constxqt left 10:53 sena_kun left 11:00 constxqt joined 11:02 jpn joined 11:04 constxqt left 11:09 jpn left 11:35 constxqt joined, Sgeo left
elcaro avuserow: Already using IPv4. Doesn't even connect with IPv6. Dunno if it's related, but I've tried from 3 different boxes, different ISP's, but all in Australia 11:38
11:40 constxqt left
elcaro If it's any help, trying to download with curl also fails with the error: 'curl: (92) HTTP/2 stream 0 was not closed cleanly: INTERNAL_ERROR (err 2)' 11:44
kjp wget also fails from here, also Australia 11:45
elcaro Looking up that error, first SO result suggests it could be an issue with 'proxy_buffering' in nginx, or maybe Application Load Balance (ALB) on AWS. 11:47
12:01 abraxxa-home joined 12:09 constxqt joined 12:14 constxqt left 12:23 lichtkind_ joined 12:26 lichtkind__ left 12:32 abraxxa-home left 12:33 abraxxa-home joined 12:35 constxqt joined 12:44 constxqt left 12:50 constxqt joined 12:59 japhb joined 13:04 japhb left 13:19 jpn joined 13:24 jpn left 14:10 japhb joined 14:17 japhb left 14:28 vrurg_ is now known as vrurg 14:43 jpn joined 14:50 sena_kun joined 15:29 jpn left 16:05 jpn joined 16:10 jpn left 16:50 jpn joined 16:55 jpn left 16:58 Sgeo joined
avuserow More data points on (pre-)compilation speed: if I make an empty A.rakumod and a B.rakumod that imports A, and change A.rakumod (to force recompilation), `use B` takes ~500ms longer than `use A`. With RAKUDO_NO_PRECOMPILATION=1, then it's only a ~18ms difference. This is on 2023.11 on older hardware; on current hardware all the numbers are roughly half as much 17:10
It seems like there is may be some potential optimization opportunities. I see that github.com/rakudo/rakudo/issues/4845 might be potentially related, but I'll file a new ticket and cross-reference. 17:11
ab5tract avuserow++ 17:12
avuserow Once I get to actual code, the 500ms difference between the A and B scenarios seems to increase. In my actual application it's more like 1.5s added. I have a few refactoring opportunities that would reduce the amount of work that compilation has to do (depending on what changed), but not until this is improved 17:15
17:32 jpn joined 17:39 jpn left 18:04 japhb joined 18:19 japhb left, japhb joined 18:42 jpn joined 18:43 japhb left 18:47 jpn left 18:48 japhb joined 18:56 japhb left 18:57 japhb joined 18:59 jpn joined 19:18 jpn left
avuserow I think a cause of slowness in precomp might be process startup time since it uses additional rakudo processes to compile. `raku -e'{}'` is 40% of the additional time between A and B above, on both computers. Unfortunately the profiling options don't seem to matter for the inner processes (and I don't even know if it's something that can be profiled?) 19:52
20:02 melezhik joined
melezhik sparrowci build with PAKKU_VERBOSE=debug looks a bit weird for me - ci.sparrowhub.io/report/3824 20:03
20:10 melezhik left 21:48 sena_kun left 21:49 sena_kun joined 21:54 rakuteen joined 21:56 rakuteen left 22:23 AlexDaniel joined
AlexDaniel I've never seen 3 almost identical PRs for the same module XD github.com/Util/Perl6-Math-Quaternion/pulls 22:23
the same issue I mean 22:24
antononcube weekly:rakuforprediction.wordpress.com/20...-via-raku/
notable6 antononcube, Noted! (weekly)
22:56 sena_kun left
avuserow I have a proof of concept of an environment variable that lets Raku use existing compiled sources but not generate new ones. This makes reloading the code much faster when changing one dependent module. It's 2x faster when changing and reloading my actual application code. 23:01
My idea is that in dev mode, I would precompile everything, then start the service with NO_NEW_PRECOMP, then run precomp in the background for next time. This should improve timing of restarting the service for developers at the cost of doing redundant compilation work. 23:02
23:25 AlexDaniel left
ugexe In process precomp is a known missing and welcome feature fwiw 23:29