01:02 tjr left, tjr joined 01:24 Manifest0 left
ab5tract tbrowder: afaik it already exists in core 01:52
In the linked module
But for some reason we don’t believe that it’s ok to let users have access to these functions 01:53
Despite providing every means necessary required to ignore or override the provided I
implementations
See my earlier grumbling and lizmat’s response 01:54
02:05 japhb left 02:09 japhb joined, yewscion joined 02:12 yewscion left
guifa anton: is there a way to do interactive chat with WWW::LLaMA? I've gotten it all set up now to get responses to prompts, but not clear how to control the conversation flow vs reseting 02:21
or is it just that the prompt needs to grow with old prompt ~ response ~ new prompt ?
02:47 kylese left, kylese joined 03:09 MasterDuke joined 03:15 kylese left, kylese joined 04:02 zara left 04:08 zetaaaa left 05:31 guifa left, MasterDuke left 05:34 Aedil joined
antononcube @guifa That can be done with a chat object (from “LLM::Functions”) or with a chatbook using magic cells. 06:10
For the former see : github.com/antononcube/Raku-LLM-Fu...onses.raku 06:12
For the latter , using a chatbook : there are several videos demonstrating LLM persona setup and chatting with it (via magic cells.) 06:18
I am not sure what is your LLaMA setup. Do you use chatbooks? 06:20
08:27 Sgeo left 09:02 hellwolf joined 09:03 teatime joined 09:04 teatwo left 09:16 hellwolf left, hellwolf joined 09:23 Manifest0 joined 09:29 sena_kun joined, Manifest0 left 09:30 demanuel joined 09:34 Manifest0 joined 09:39 Manifest0 left
demanuel quit 09:39
09:39 demanuel left 12:17 guifa joined
antononcube quiet 12:37
12:59 rba left 13:02 rba joined 13:42 teatwo joined
tbrowder ab5tract: didn't you mean patrickb ? 13:43
13:45 teatime left
tbrowder i guess someone could just transform that internal test to a published module... 13:47
ab5tract tbrowder: irclogs.raku.org/raku/2024-11-11.html#16:56 13:52
irclogs.raku.org/raku/2024-11-11.html#17:16
There should be no reason to duplicate working code
But others do not agree 13:53
tbrowder ok, i did see that, but in my brain fog i basically saw "fwiw, i don't either" at least at one point--anyhow, i understand a bit about reluctance having been a part if having some things accepted to add to core but more not. 14:09
the test thing i dunno...
*of having... 14:10
i am very happy with raku and wish we could get more traction with a wider user set 14:11
change of subject: perl and raku foundation. i was disappointed in the remarks by the board member about how deeply in debt they are and the main source of funds is from the conferences. is that true? 14:16
i can tell you i have attempted correspondence with the board several times over the years and gotten no response. in the us there are megabucks being given as grants that could be had with a little support. 14:19
[Coke] trying to track down ToddAndMargo's Terminal::ANSIColor issue on windows - getting an error "failed to update fez mirror: 360.zef.pm" on my windows box. 14:22
(on `zef update`0 14:25
if I go to that URL in chrome, opens fine. 14:26
is there a way to get verbose output about that failure? 14:27
ab5tract [Coke]: maybe `zef --verbose` ? 14:29
of course you can also bypass that issue by cloning the repo directly, but I don't think this is news to you :) 14:30
[Coke] Yah, it's just a PITA on windows to do that. 14:31
(well, slightly more of one)
ah, curl 360.zef.pm/ dies with CRYPT_E_NO_REVOCATION_CHECK 14:34
probably need to add that self-signed cert to the OS 14:46
patrickb Try disabling the AntiVirus solution. 14:48
14:49 hellwolf left, hellwolf joined
patrickb 360.zef.pm does not use a self signed cert. 14:49
[Coke] This is a corporate laptop, that's a no go 14:54
ok. then just "that cert"
patrickb 👍🏼 14:55
[Coke] importing the cert to windows... did nothing for the curl error 15:10
I'm still seeing a self signed cert in cert chain in the openssl output. digging... 15:14
ab5tract hmm, it doesn't seem likely since it is working in Chrome but sometimes I chase my tail in similar situations only to realize that the clock is wrong 15:22
[Coke] Guessing this is because I'm on a corporate laptop that probably is intercepting a bunch of stuff on my behalf. 15:32
patrickb I guess curl / zef are not using the Windows cert store, but the one curl brings along. And the cert that the AV installs to introspect the traffic is only installed in the Windows store.
[Coke] if I could find a way to disable the revocation check, that would probably be fine and safe here. 15:33
patrickb Does your windows bring a curl with it? 15:38
If so, which curl are you using in your tests?
If you can modify the PATH, so it's using the Windows provided curl, that might possibly fix both curl and zef 15:39
What's the output of `where curl`? 15:43
[Coke] cmd doesn't have a where, one sec. 15:52
ah, it does have where, which I didn't know until just now - C:\windows\system32\curl.xe 15:53
ah, it does have where, which I didn't know until just now - C:\windows\system32\curl.exe
I meant, it doesn't have `which`, which I auto-translated your ask into in my brain. whoops
ab5tract [Coke]: if you can run powershell instead of cmd, I'd expect it to save you some sanity 15:57
[Coke] I switch back and forth. also have git bash 15:58
16:00 yewscion joined
ab5tract yesterday I learned that there is an option when install git bash that one can specify that bypasses the whole is-no-true-TTY issue 16:00
apparently you can click a checkbox that makes it wrap around either cmd or powershell instead of the GNU option 16:01
patrickb Ok. Then it's already using the windows native curl. Then my idea was a red hering. 16:03
[Coke] trying answers.microsoft.com/en-us/window...e29d54c02e 16:09
\p/ 16:11
er, \o/
though I guess I have a neck in that first emoticon.
... damn. that fixes *curl*. it doesn't fix zef. 16:12
AHH. no. I tested on cmd pre-reboot, and powershell post-reboot. it changed nothing. :| 16:13
(zef fails in powershell also, even though curl direct works)
can't install zef HEAD on my box with 2024.07+ 16:18
error while compiling zef\CLI.rakumod at CLI.rakumod line 2
... updating raku...
16:21 yewscion left
[Coke] src\moar.h "cannot open inclue file dynload.h" 16:22
guessing so far this is just my old windows install needing to be nuked and done over. :| 16:24
ab5tract [Coke]: hmm, perhaps... though the compilation story is actually better on older Windows if IIRC? 16:25
which compiler are you using?
[Coke] cl reports 19.29.30139 16:26
ab5tract msvc?
[Coke] (in the x64 native tools command prompt)
y
doing a fresh build with 2024.10 rakudo tag 16:28
nuked nqp/moarvm source
... I was just trying to answer toddandmargo's ansi question. :) 16:29
ab5tract :)
hmm, I don't recall the "compiler cutoff" 16:30
[Coke] funny error: "macro expansion producing 'defined' has undefined behavior"
ab5tract but it *might* be 2018, where your version seems to be 2019
[Coke] I've used 2019 before. 16:31
ab5tract ok
[Coke] I've never updated it on this PC.
I just haven't built on windows in months; (I did upgrade to win11 at some point)
ab5tract the win11 upgrade is probably not an issue... 16:32
but I could not get a working compilation starting already in Windows ~2-3 years ago
[Coke] I do wonder why it's using a lib in program files (x86) when I'm in the x64 build cmd.exe 16:33
ab5tract that's definitely suspicious 16:34
[Coke] we seem to have gotten further after nuking nqp/moar and using the release tag.
ab5tract or rather, that definitely *feels* suspicious
16:37 yewscion joined
[Coke] yah, into nqp now 16:37
... and now strawberry perl.exe is dying "return code 0xff' after the stage start: 0.000 output 16:42
(in rakudo biuld)
ab5tract ufff 16:45
Geth ecosystem/main: eb0a140abd | (Elizabeth Mattijsen)++ | META.list
Removing masak's modules

These will not be updated anymore, and will therefore remain available through the REA		 16:58
17:06 hellwolf left
[Coke] "C:\Strawberry\perl\bin\perl.exe" rakudo-m-early-build --setting=NULL.c --ll-exception --optimize=3 --target=mbc --stagestats --output=blib\CORE.c.setting.moarvm "gen\moar\CORE.c.setting" 17:07
17:07 hellwolf joined 17:08 zetaaaa joined
[Coke] why is rakudo-m-early-build using the *installed* nqp path? 17:09
it's got the local rakudo lib, but the install path for NQP lib.
also, apologies for doing this in raku and not raku-dev, wasn't expecting it to turn into a build issue! 17:15
17:22 sayan39 joined
Geth ecosystem/main: 47714f8574 | (Elizabeth Mattijsen)++ | META.list
Remove atweiden's modules that won't be updated anymore

  github.com/atweiden/config-toml/is...2470973451		 17:22
17:23 sayan39 left, sayan39 joined
ab5tract Huh, that does seem strange… I assume you ran perl Configure.pl with —gen-moar and —gen-nqp 17:24
17:25 tea3po joined 17:28 teatwo left 17:32 japhb left 17:37 japhb joined 18:11 yewscion left
Geth ecosystem/main: 3c908a0604 | (Elizabeth Mattijsen)++ | META.list
Remove other atweiden's modules

soon to appear in the zef ecosystem		 18:15
18:19 yewscion joined 18:42 yewscion left 19:00 yewscion joined 19:49 Aedil left 20:01 hellwolf left 21:13 yewscion left 21:53 yewscion joined 21:58 yewscion left 23:11 yewscion joined 23:15 yewscion left, sena_kun left 23:47 Sgeo joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!