🦋 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.
antononcube @ab5tract Great! 00:11
@ab5tract No, it does not work -- see my comment in that PR. 00:39
Really thank you for trying out to fix this issue~
ab5tract No problem! It’s my fault that it exists so… :) 01:23
I’ll take a look at how jupyter is passing these arguments. What’s unfortunate is that I could see that maybe it wants the the argument as one entry and the value as the other, but raku requires an equals sign between the arg and its value 01:25
ahhh! I think I grok it now 01:34
antononcube: update pushed. can't check right now unfortunately but more details available in the PR 01:55
grondilu Sometimes I get "The following packages were stubbed but not defined:" error, sometimes I don't. It's very confusing. 03:41
hard to golf, though. 03:42
m: class A {...}; class B {...}; class A { has B $.b }; class B { has A $.a; }
camelia ( no output )
grondilu gist.github.com/grondilu/327bbf1f8...04056736fe 03:44
grondilu ^this code throws the aforementioned error. No idea what's different about it than the trivial code above. 03:47
well nevermind it seems that a closing curly brace was missing 03:52
ab5tract Hmm.. sounds like it may still be worth reporting as LTA? 04:09
grondilu probably not until I can reproduce it with a smaller code 04:36
librasteve @antononcube i remain rather confused about the interaction between Jupyter::Chatbook and Jupyter::Kernel … as set out here, github.com/antononcube/Raku-Jupyte.../issues/4, the two modules are confused by zef which takes them as the same module 08:20
it seems that fixing the aforementioned bug in Jupyter::Chatbook but not Jupyter::Kernel has the potential to increase this confusion 08:21
technically it seems the choice is (i) to just de-emphasize Jupyter::Kernel and take Jupyter::Chatbook as the new standard bearer or (ii) to refactor Jupyter::Chatbook so that it takes Jupyter::Kernel as a dependency or maybe a Github submodule 08:25
I get the feeling that (i) is now happening de facto due to your more attentive maintenance… would you agree? if so I’ll start to move my non LLM examples and Docker builds over to Jupyter::Chatbook descendant… 08:28
patrickb ab5tract: I've just read your Comma advent post 11:10
I had no idea how much work was involved to get it revitalized. Reading about it is humbling. I'm really impressed. 11:11
Thank you for all the work you put in to this!
lizmat indeed! ab5tract++ 11:13
librasteve i would 3rd that.. very interesting and comprehensive post - I've been rootling around in the Java recently (maybe add crotmp fragment syntax highlighting) - it is a surprisingly extensive - kudos to ab5tract I use this every day 11:22
Geth advent/main: d0a52531f6 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | raku-advent-2024/authors.md
swap 21 <-> 23
 11:35
ab5tract Aw shucks, thank you all for the kind words! :)
I should mention that my efforts pale in comparison to the original implementation efforts. But it has also indeed been quite some work for me too, so thanks again for sharing your feedback ❤️‍🔥 13:04
lizmat
.oO( we stand on the shoulders of giants )
but we're not too small ourselves either :-)
antononcube @librasteve I have to (re-)investigate the zef confusion between “Jupyter::Kernel” and “Jupyter::Chatbook”. 13:51
As for “Jupyter::Chatbook” being the standard bearer — maybe it is good idea, I have to think about it. 13:53
The primary reason “Jupyter::Chatbook” not to be the standard is that it has dependencies on many packages. None of them is particularly “heavy”, except “LLM::Functions” and “LLM::Prompts“. (But well, those are the ones that “make the chatbook.”) 13:55
As for refactoring — yes, the original code of “Jupyter::Kernel” had to be refactored to some extend in order to add more magic cells. And think my refactoring-level understanding of the code is in that sub-domain. 13:59
I think initially I wanted “Jupyter::Chatbook” to just depend on “Jupyter::Kernel”, not to be a fork of it. But that would have required too much “understanding” and, hence, much less LLM-ing. 14:02
(Meaning, I wanted to just see and experiment with LLM-features in a Raku-centric notebook.) 14:03
librasteve yeah - I get that ... I was thinking of a PR on Jupyter::Chatbook to put the use X Y Z list into a config setting / file 14:13
not ready to / needing to do that for a while though since I am up to my eyeballs in Cro 14:14
antononcube This would require a very streamlined inclusion of new magic cells. Similar to that in "IPython", but better. 14:15
librasteve btw it's kind of nice to have all the LLM stuff there if needed and only costs a bit of install time 14:16
antononcube That is one of the primary motivations!
librasteve so probably I'll just broaden my use of J::C as it is and pre-build some Docker images
antononcube Then I included MeraidInk, DeepL, and Wolfram|Alpha. 14:17
Again, most these are relatively simple "WWW::*" packages.
librasteve cool 14:18
antononcube Except ""WWW::OpenAI" and "WWW::Gemini" -- have certain complicated logic.
The most useful new feature of "Jupyter::Chatbook" -- related to what you suggest above -- is to have configurable LLM-chat personas. With, say, JSON file. (Or TOML or YAML.) 14:19
We have discussed that before...
I might implement that before- or around the New Year. 14:20
ab5tract antononcube: looks like you are getting a free Dockerfile revision with the repl fix :) 14:24
it looks like the `rakudo/daily` images stopped being updated early last year 14:25
antononcube 🙂 yeah, that would be nice.
ab5tract antononcube: do Jupyter::Chatbook tests pass cleanly for you with an older rakudo version? Because it seems that t/02-sandbox.t is failing for me with and without my changes 15:10
antononcube Yes, they do. I have re-installed "Jupyter::Chatbook" multiple times for moar-2024.08 and moar-2024.09 while trying out the fixes. 15:46
ab5tract ++ 15:58
ab5tract I was hoping to fix it without any changes to Rakudo, but it looks like that's a requirement 16:22
but there are still the free bonus Dockerfile adjustments :) 16:23
antononcube 👍
ab5tract antononcube: rakudo.git @ 1bd183d400 should address the issue 16:42
linkable6 (2024-12-20) github.com/rakudo/rakudo/commit/1bd183d400 Pick up repl-mode default fix in NQP
ab5tract now fully tested thanks to a new Dockerfile that builds rakudo from scratch 17:19
antononcube Ok, great! I will try it out soon, today. 18:46
Geth ecosystem/main: 4f921ad057 | (Elizabeth Mattijsen)++ | META.list
Remove Business::CreditCard

It lives on as a Raku Community module now		 19:04
ecosystem/main: ef02d91c44 | (Elizabeth Mattijsen)++ | META.list
Remove Cache::LRU

This has basically been superseded by Hash::LRU		 19:48
ecosystem/main: 8f2cceea6e | (Elizabeth Mattijsen)++ | META.list
Removed WebService::Google::PageRank

The webservice used appears to have bitrotted		 19:57

Powered by Cro and the Raku® Programming Language.

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

Thank you!