00:51 melezhik_ joined 00:53 melezhik_ left 01:52 melezhik_ joined 01:54 melezhik_ left 01:58 librasteve_ left 02:53 melezhik_ joined 02:57 melezhik_ left 03:54 melezhik_ joined 03:57 melezhik_ left 05:30 melezhik_ joined 05:52 melezhik_ left 06:09 melezhik_ joined
melezhik_ o/ 06:09
postmortem - o10r stopped yesterday , no disk space , it managed however finish 664 modules in 1 hour, which is approximately around 3 hours for 2000 modules 06:11
and again this us with sub optimal version of patrickb: biggest agent (19 threads), had it been the latest 0.0.28 results would have been even better, but even this time is good already 06:16
this is another example when dependencies installed during module installation are marked as processed, to exclude them from list to test - gist.github.com/melezhik/df023e66a...9d8e7cec8e 06:20
zef install Calendar, install some deps, see `add module` lines in the report
for example - `add module JSON::Fast` ... 06:21
06:23 melezhik_ left 06:29 melezhik_ joined 07:08 melezhik_ left 07:15 melezhik_ joined 07:17 melezhik_ left 08:22 melezhik joined
melezhik . 08:22
ab5tract: [Coke] patrickb please upgrade agent from the latest commit and restart it. Thanks . Hopefully this time won’t be any o10r crashes )) 08:46
After this upgrade agent upgrades could be performed without container rebuild and restart 08:50
github.com/melezhik/brownie/tree/m...-container 08:51
11:05 melezhik_ joined 11:17 melezhik_ left 11:21 melezhik_ joined 11:22 melezhik left 11:25 melezhik_ left 11:37 melezhik_ joined 11:57 melezhik_ left 12:14 melezhik_ joined 12:23 melezhik_ left, melezhik_ joined 12:39 melezhik_ left 12:40 melezhik_ joined 12:47 melezhik_ left 12:48 melezhik_ joined 12:53 melezhik_ left, melezhik_ joined
melezhik_ . 13:03
13:14 melezhik_ left, melezhik_ joined 13:17 melezhik_ left, MasterDuke joined
MasterDuke patrickb: t/04-nativecall/26-varargs.t passes on my branch on my x86-64 linux desktop. so the warning on my aarch64 linux laptop may in fact be pointing out the problem 13:19
tellable6 2025-11-26T21:38:07Z #raku-dev <patrickb> MasterDuke: Just did another quick test run locally. All green.
MasterDuke and yeah, t/04-nativecall/26-varargs.c:64 from `char *a1 = va_arg(args, char*);` to `signed char *a1 = va_arg(args, signed char*);` gets it passing 13:21
oops, line 54 13:22
Geth rakudo: MasterDuke17++ created pull request #6033:
Explicitly use signed char when comparing to -1
13:27
timo quick question, do we have aliases inside NativeCall for both signed char, unsigned char, and "whatever just `char` means on this system"? 13:29
MasterDuke i don't think we have char at all in NativeCall 13:31
just Str, which i guess maps to "whatever just `char` means on this system" 13:33
13:34 melezhik_ joined
MasterDuke but patrickb would know better than i. i've never used NativeCall, at least not directly 13:34
timo Str has some magic handling that makes sure we encode when going to C and decode when coming back from C 13:35
MasterDuke ah, right
lizmat: on a different topic, did you see www.theregister.com/2025/09/30/cyb...on_column/ ? might be of interest 13:38
lizmat MasterDuke: thanks for the link 13:39
am mostly afk at the moment, but maybe sjn has some ideas about it: www.theregister.com/2025/09/30/cyb...on_column/
timo which is where the decision to use char vs wchar or bigger would come from, by choosing a different encoding (though since C ABI doesn't differentiate between char* and wchar* i'm not sure we bother to do anything different. i think encodings are defined over byte arrays, not arrays of larger-than-byte items?) 13:40
melezhik_ finally agent upgrade is just a `docker exec -it agent bash /opt/brownie/agent/upgrade.sh` 13:45
sjn reads
oh yeah, that one
sjn disagrees with gregkh 13:46
well with some of what he says
13:54 melezhik_ left 13:58 melezhik_ joined
MasterDuke afk for a bit 14:04
patrickb ugh, I didn't know signedness could also differ. I thought only size was allowed to differ. 14:07
14:08 MasterDuke left
patrickb hi 14:08
melezhik_: can you priv msg me again? my client fails to open the window atm. another ping could help... 14:12
14:18 melezhik_ left 14:19 melezhik_ joined 14:24 melezhik_ left 14:25 melezhik_ joined
melezhik_ done 14:25
14:39 melezhik_ left
Geth rakudo/main: 7a302e082c | MasterDuke17++ (committed using GitHub Web editor) | t/04-nativecall/26-varargs.c
Explicitly use signed char when comparing to -1

Some systems (e.g., Arm) default to char being unsigned.
15:06
15:43 melezhik_ joined 15:44 melezhik_ left 15:54 melezhik joined
melezhik . 15:54
[Coke] I'm on the road for a bit, no agent for a day or so
disbot6 <melezhik.> Coke - no worries, just keep you updated on overall progress , thanks 16:05
16:39 guifa joined
melezhik . 17:18
17:59 finanalyst left 18:06 librasteve_ joined 19:21 guifa left 20:14 guifa joined 20:19 guifa left 20:23 melezhik left 20:24 melezhik joined 22:33 melezhik left