00:04
jaguart joined
|
|||
antononcube | Rakunista, Rakultist, Rakutopist. | 00:07 | |
00:08
jaguart left
|
|||
[Coke] | muckrakers | 00:13 | |
drakes | |||
00:18
wayland left
00:37
MasterDuke joined
|
|||
timo | ... Rakuna Matata ... | 00:51 | |
01:02
wayland joined
|
|||
ingy | librasteve: $ ys -pe 'ENV.keys().filter(\(/XDG/)).take(5).join(",")' | 01:13 | |
"XDG_SESSION_DESKTOP,XDG_CONFIG_DIRS,XDG_CURRENT_DESKTOP,XDG_SESSION_TYPE,XDG_RUNTIME_DIR" | |||
like that? ys has the env var mapping in `ENV` | 01:14 | ||
antononcube: sorry I didn't reply the other day. was about to fall asleep and wasn't sure exactly what you were asking (iirc) | 01:15 | ||
_grenzo | Rakumedians, Rakuligans, Rakumancers | 01:16 | |
01:17
hulk joined
|
|||
ingy | wielders of Rakudoo | 01:17 | |
Rakutube? | |||
01:17
kylese left
01:18
Manifest0 left
|
|||
_grenzo | Rakunnovators | 01:19 | |
Rakuthiasts, Rakumaticians | 01:24 | ||
wayland | Do we have anything amongst the LLM-related modules that assists with RAG (specifically, I'm going to want to edit source code using the LLM) | 01:35 | |
02:01
Sevalecan left
02:04
Sevalecan joined
02:15
hulk left,
kylese joined
02:30
kylese left
02:34
kylese joined
02:51
andydude joined
|
|||
andydude | Hey, a little bird told me that my package can no longer be updated and that I need to switch to zef | 02:52 | |
[Coke] | probably was 'fez' | ||
andydude | Could be, I have no idea | ||
github.com/andydude/p6-c-parser | 02:53 | ||
This is the package | |||
How would I go about fezzifying it? | 02:54 | ||
MasterDuke | does lizmat have a tutorial for converting a dist to fez? | 03:01 | |
also, tonyo1 would be a good person to ask | |||
andydude | also, should I rename it raku-c-parser? | 03:03 | |
coleman | Yea, and github will redirect old links :) | 03:05 | |
03:26
Aedil joined
05:50
wayland left,
andydude left
05:53
wayland joined
06:04
jaguart joined
06:09
jaguart left
|
|||
antononcube | @timo Obligatory eating of bugs by everyone is coming soon… | 06:19 | |
@wayland It is in my TODO list to make RAG-related modules/packages. The procedure is relatively simple and have done a few times. I am mostly “blocked’ by appropriate naming of the package(s) and functions/subs in them. | 06:22 | ||
@ingy My question was about converting Markdown to YAML — it is possible to a point, but there is some “information loss.” | 06:26 | ||
@wayland You can edit source using LLM with the currently available packages. 🙂 Interactivity is key — that is why chatbooks were made. The editing can be done the subs of “LLM::Functions”. I would say you do not need RAG at all. | 06:35 | ||
I.e. RAF can be used to locate certain code chunks in large code bases. But with the most recent models that take 100s to thousands of tokens (or even millions) that is not that needed. | 06:37 | ||
Of course, if you want to use locally run models — for example LLaMA models — that take and produce relatively small number of tokens, then yes, RAG becomes more useful. | 06:39 | ||
There are several packages for dealing with “vector databases” produced by RAG procedures: “Math::Nearest” and related K-D tree packages, and “ML::StreamsBlendingRecommender”. | 06:46 | ||
The latter uses (currently) camelCase — that is why it is not published in the Zef ecosystem. | 06:48 | ||
wayland | antononcube: Great! Good info, thanks! Will likely be using LLM::* as produced by you :) | 06:50 | |
antononcube: I've been meaning to ask, do your reshapers write to data sources, or are they only for querying? | 07:29 | ||
.tell librasteve A difference between TOP and what I see in a quick skim over of R dataframes is that I'm planning to try do a better job of allowing for more spreadsheet-like layouts (column typing optional, but can maybe have formulas in cells). But maybe that's just my ignorance of dataframes | 07:32 | ||
tellable6 | wayland, I'll pass your message to librasteve | ||
antononcube | @wayland “Data::Reshapers” is for Raku data structures only. It uses the concept of dataset: array of hashes. | 07:37 | |
07:39
wayland76 joined
07:40
wayland left
|
|||
My primary focus on data wrangling is facilitating the execution of the same workflows over different languages and libraries. So, I often use “DSL::English::DataQueryWorkflows” — that can produce code for writing to data sources. | 07:41 | ||
But since I take functional programming’s perspective on data wrangling there is minimal support for such operations . | 07:42 | ||
wayland76 | antononcube: OK, great. I just looked at the examples and didn't see anything for writing to data sources :) | 07:43 | |
(examples+doco) | |||
antononcube | My perspective is minimalistic— there is no need to implement or address certain operations since they are part of the Raku language. | 07:52 | |
Here is long explanation / overview: youtu.be/efRHfjYebs4?si=9TgqnnCShk...amp;t=1294 | 07:53 | ||
wayland76 | Sure :). It all depends whether you want to prevent questions like this one :p | ||
Thanks! | 07:55 | ||
Loaded in a browser tab so I can watch it later :) | |||
07:58
sena_kun joined
|
|||
antononcube | @wayland Again — my primary efforts in Raku for data wrangling were/are to produce code for other languages and/or their corresponding libraries. I started working on facilitating Data wrangling with Raku much later. | 07:59 | |
wayland76 | antononcube: Good context -- appreciated. | ||
08:01
wayland76 left
08:04
wayland76 joined
08:09
sena_kun left
08:49
bdju left
08:51
bdju joined
|
|||
ab5tract | To clarify, the singular of Rakuminati is a Rakuminatus | 09:11 | |
But it’s more likely that one would use the plural. “Oh, tbrowder is Rakuminati for sure” | 09:12 | ||
I think it’s important to feature the fun in having names like these, so outside of the question of what the flair is on reddit, I think the most appropriate approach is that everyone uses their favorites as they feel like it, which could change every sentence if so desired | 09:14 | ||
lizmat | weekly: news.ycombinator.com/item?id=41470958 # more complex than it can be | 09:20 | |
notable6__ | lizmat, Noted! (weekly) | ||
ab5tract | That’s a sad review :( | 09:25 | |
lizmat | I feel I could write a similar review about Python in the few hours that this person spent on Raku | 09:29 | |
ab5tract | Yeah.. I wish there was actionable feedback here as well. What part of specifying dependencies is particularly troublesome, for example? | 09:30 | |
librasteve | that’s looking good - and if I run ys the YAMLScript at the start of my Raku code then can YS write to the ENV in that process? | 09:32 | |
tellable6 | hey librasteve, you have a message: gist.github.com/b5533e835446c50bda...3356b15f9c | ||
09:44
Aedil left,
itaipu left,
Altreus left,
tib_ left
|
|||
wayland76 | ab5tract: Regarding Rakuminatus, thanks! :) | 09:46 | |
09:50
tib_ joined,
Altreus joined
09:51
itaipu joined
10:04
Aedil joined,
jaguart joined
10:08
jaguart left
11:03
Sgeo left
12:13
Manifest0 joined
|
|||
antononcube | Rakuminatus - Rakumishmatus. | 12:35 | |
Also, “Camelia” derivatives — e.g. “Camelinati” — can be interpreted as some sort political preferences. (Vice-president Harris,) | 12:37 | ||
Well, in USA at least… | |||
tbrowder | ref bad raku review somewhere above: i'm confused, all i see is the short one by librasteve which looks very positive to me | 13:28 | |
although i disagree about ease of learning python for those of us with short memories | 13:32 | ||
timo | the original comment librasteve was responding to is not visible any more, instead it just shows "[flagged]", i can't even go to its parent node? | 13:33 | |
tbrowder | gotcha, thnx | 13:34 | |
antononcube | This kind of censorship is annoying and disrespectful -- it assumes we cannot judge criticism, good or bad, ourselves. | 13:46 | |
13:58
sena_kun joined
|
|||
timo | there's a difference between censorship and moderation. i don't know what the original text was, so i can't make a judgement on whether it was reasonable to nuke that comment, but you also want to, as much as possible, prevent your community from becoming a festering cesspool of negativity and insults | 14:35 | |
antononcube | @timo It seems you have internalized some of the censorship mantras. | 14:39 | |
timo | *shrug* | 14:40 | |
antononcube | I do not disagree with your statement, though. | 14:41 | |
timo | there is definitely a gradient, at one end of which sits the most obvious spam that you definitely want to keep out, somewhere on that gradient is criticism of different quality levels and different levels of aggressiveness | 14:44 | |
i assume that user was free to rethink and reword their comment and put it back in the same thread if it meets some minimum of adhering to whatever guidelines are in place | 14:45 | ||
i also find it important to remember, if you kick out obnoxious people, you immediately get to hear from obnoxious people and people who don't mind them so much, but if you leave in obnoxious people you rarely get to hear from other people who decide they don't have stick around and put up with the obnoxious people | 14:50 | ||
and putting the onus of muting/blocking obnoxious people on x times as many non-obnoxious people seems like a very weak fix | 14:51 | ||
reddit of course has the upvotes and downvotes to make comments that don't add to a discussion trend to the bottom and become hidden by default, but anecdotally i see often enough what looks like people disagreeing with a comment downvoting it | 14:53 | ||
i barely read The Orange Site so i can't really opine on that | |||
15:37
jgaz joined
15:38
jgaz left
15:42
jgaz joined
|
|||
librasteve | btw i’m as mystified as you guys what happened to the original … since i’m in europe i got the heads up and thought i would make some rapid comment since early comments tend to stick at the top of hn threads and i wanted some kind of rebuttal / balancer in case the topic got big traction | 15:53 | |
my guess is that the OP got cold feet and pulled the comment or that the mods pulled it as too -ve. ??? | 15:54 | ||
@timo i agree with you point that weak moderation promotes silent leavers … even if one takes up the argument of the troll, others get tired of watching the ding ding | 15:56 | ||
ding dong | |||
antononcube | @librasteve Next time take screenshot of the text you comment on! | 16:32 | |
Or at least, just the controversial ones. | |||
@wayland Here is a dedicated Raku-RAG respository: github.com/antononcube/Raku-LLM-Re...Generation | 17:33 | ||
holmdunc | I don't think it was flagged for negativity, more like because that type of submission is supposed to ask the community a question and not just dump an unsolicited opinion | 18:06 | |
cdn.discordapp.com/attachments/633...99a9c& | |||
18:10
andydude joined
|
|||
librasteve | btw my toolbox comments was inspired by @wayland … it was the best thing i could think of in a service station on the M5 … sorry for stealing it | 18:21 | |
scullucs | "I am a beginner and things aren't immediately and effortlessly obvious." | 18:27 | |
19:04
bdju left
19:14
bdju joined
|
|||
ugexe | I don’t think their gripe is related to being a beginner. Anyone who wants a simple language should arguably share that opinion regardless of their experience level (and maybe even more so for experienced developers) | 19:19 | |
scullucs | I guess what I meant is more "I spent a couple of minutes on a language I know nothing about and ...". | 19:23 | |
But I agree that Raku is not a simple language. | 19:27 | ||
ugexe | I don’t get that impression myself. They are aware about the lack of TCO and point out other details that suggest they spent some time on the details someone just banging out a program probably wouldn’t have | 19:28 | |
Nothing they said was particularly profound mind you. It just isn’t the right language for that type of developer | 19:31 | ||
scullucs | I guess you're right. I think I got annoyed by the "multiple ways to do one thing is confusing" and being surprised at "having to reread the docs to get familiar with the syntax". | 19:32 | |
19:37
jgaz left
|
|||
librasteve | When I learned raku (perl6) it was coming back into coding after several years break. I employed Think Perl6 greenteapress.com/wp/think-perl-6/ which is based on Think Python. This graduated approach gets you to a "strong beginner" level that combines the imperative, functional, oo and grammar aspects of raku. The newer equivalent imo is raku.guide. I like to think that there is an "easy raku" | 19:54 | |
level that can be used by coders who just want the Python level of depth. And, in general, even expert rakuteers are writing code that can be read and fixed by less proficient colleagues - ie just use the level of tooling needed for the job, don't obfuscate. So raku can be easy and it can be hard in the tradition of the easy things are easy and the hard things are possible. | |||
That said, raku is truly large and layered and it is easy to wander off the narrow path of "easy raku". Anyway these are just idle musings. | 19:56 | ||
[Coke] | iwbni gist.github.com let you delete old gists from the "all gists" page. | 20:19 | |
20:22
Aedil left
20:38
andydude left
20:50
Sgeo joined
22:03
sena_kun left
|
|||
tbrowder | for you maths xperts i just found a reference you might find interesting: <pomax.github.io> was looking for good poop on bezier curves for input for pdf graphics operators | 22:15 | |
look for his primer on bezier curves | 22:18 |