🦋 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.
01:26 kylese left 01:27 kylese joined 01:44 dg left 01:45 kylese left 01:49 dg joined 01:56 dg left 02:02 dg joined 03:14 Aedil joined
ab5tract antononcube: Nice article! I stumbled on this code though: $poet4.defined && $poet4 ?? $poet4 !! $poet3 07:43
Why not $poet4 || $poet3 ? 07:44
Voldenet more expressive 08:22
same reason why people prefer True and False over !0 and !1 despite them being shorter 08:32
08:40 kylese joined 09:07 Aedil left 09:10 oodani left
jubilatious1_98524 In the R-programming language that's written: if(!is.na(poet4)) ifelse(test=poet4, yes=poet4, no=poet3) 09:14
Or more simply (because R understands positionals if named-arguments are not used): if(!is.na(poet4)) ifelse(poet4, poet4, poet3) 09:15
09:16 oodani joined 09:51 Sgeo left
ab5tract Voldenet: it reads as overly verbose and pedantic. I don’t agree at all with the comparison to inverted integers standing as boolean 10:22
And besides, I was asking the author 10:23
I guess through acts of borderline conceptual violence, someone could create an undefined object which is also True 10:25
Writing overly verbose code to protect against corner cases like that is what I disliked most about Perl, fwiw 10:26
10:27 Aedil joined
ab5tract jubilatious1_98524: that’s how now::if works (minus the named params). It can be pretty handy to have that construct, even if it can also feel a bit clunky as a bit of syntax 10:43
10:46 cryosis left 10:47 cryosis joined, cryosis left, cryosis joined
jubilatious1_98524 @ab5tract I'm just coming back to R so a bit rusty. Not sure it can be simplified and wheter-or-not that's what @antononcube intends. R used tri-valued logic so poet4 = NA works. But automatic char -> logical conversion doesn`t. 10:52
ab5tract the || operator ($a || $b) is literally doing $a.defined && $a.so ?? $a !! $b 10:55
Again unless user space has introduced objects that not defined but are true 10:56
So I’m not sure there’s any need for a new construct 10:57
10:57 cryosis left
ab5tract But it’s accessible via use nqp if it’s wanted 10:58
10:59 cryosis joined
ab5tract sees nqp::if was autocorrected to now::if in a previous message :/ 10:59
jubilatious1_98524 AboveR code might need isTRUE(x), which in R "is the same as { is.logical(x) && length(x) == 1 && !is.na(x) && x }; isFALSE() is defined analogously. 11:00
Thx, nqp:if makes sense.
I think I understand. In R you can define x <- NA and you'll get back is.na(x) #TRUE, but if you haven't defined y you'll get back is.na(y) #Error: object 'y' not found 11:04
But you're saying you can declare an object tyoe in Raku which is True. 11:05
'tyoe' should be 'type' 11:06
(cool, editing works).
I'm glad Raku has True and False, in contrast someone told me Perl started with !!1 and !!0. 11:42
Voldenet maybe I went overboard with !0 and !1, but it's a neat trick, I'll use it ;) 11:43
I wasn't explaining, just trying to justify the "why" behind it 11:46
I do like expressive syntax though, even if it means verbose code 11:48
so I like things like `if(thing == false) { }` 11:49
even if they're not elegant and redundant at times
> Can't locate object method "say" via package "true" (perhaps you forgot to load "true"?) at -e line 1. 11:52
going back to perl after using raku for a while is not pleasant 11:56
13:20 wayland76 joined, wayland left 13:51 human-blip left 13:53 human-blip joined 14:05 LainIwakura joined
antononcube @ab5tract Good point. Basically, that was an initial “LLM::Graph” development and documentation example. “LLM::Graph” provides asynchronous (LLM-)programming and initially I was not sure how reliably the input variables to the graph are propagated and treated with the introspection routines. 14:18
For example, any Whatever before given to the LLM-functions is replaced with the empty string, etc. 14:19
Also, as a documentation example I want to be seen that $poet4 can be undefined -- this hints that corresponding sub is not always evaluated. (Because of the result of the corresponding test function.) 14:28
14:34 LainIwakura left
ab5tract antononcube: makes sense! 14:41
It’s always a balancing act with this stuff. My reaction was that spelling it out this way may give the impression that Raku requires that kind of detailed style of handling defined truthies 14:44
Voldene: fair point. I reacted a bit quickly there, should have processed a bit I think 14:45
WRT Perl, I think ‘use v5.38’ should give you say (and sub signatures, with a new enough version). 14:49
‘say’ was always a tough one for them to put into the namespace because I guess there were tons of bespoke implementations on the DarkPAN
antononcube I am not sure now -- now I think therer should be a remark in the post that clarifies the verbose code. Or the other way around uses the short code and explains that $poet4 can be undefined, but also be an empty string. 14:52
Currently, "LLM::Graph" does not handle broken promises directly, and it should. So far, LLMs handle those nicely, but the handling should be more targeted.
ab5tract antononcube: maybe another option to highlight these potential states would be to use a named subset 15:08
15:24 melezhik joined
melezhik . 15:24
15:35 Aedil left 15:45 Aedil joined 17:18 jgaz joined 17:34 melezhik left 17:57 librasteve_ joined
librasteve I just fezzed github.com/librasteve/FatRatStr which solves the issues covered in github.com/issues/created?issue=Ra...ving%7C494 ... ideally this would have used class FatRatStr is Allomorph is FatRat {} but I couldn't get the inheritance to work ... anyone with proper raku skills welcome to come in and advise ;-) 18:18
also ... perhaps a candidate for adding to rakudo core (?) 18:19
18:28 eseyman left 18:32 manu_ joined, samebchase left 18:33 samebchase joined
SmokeMachine AI has helped me to create new .groups for ASTQuery, does anyone have any opinion about that? Does anyone have any suggestion on new groups I should add? github.com/FCO/ASTQuery/blob/main/REFERENCE.md 18:41
19:06 Sgeo joined 19:17 TempIRCLogger joined
lizmat . 19:17
19:18 TempIRCLogger left 19:21 TempIRCLogger joined
lizmat TempIRCLogger: Hello in there 19:21
19:21 TempIRCLogger left, TempIRCLogger joined 19:23 TempIRCLogger left, TempIRCLogger joined 19:24 TempIRCLogger left 19:25 TempIRCLogger joined
librasteve hello 19:36
lizmat PSA: it looks like irclogs.raku.org and Geth will be down on Mon 15 Sep from 0600 - 0800 UTC due to electricity works at their data center 19:39
twas in preparation for ^^ :-)
librasteve SmokeMachine: gosh ... that looks very comprehensive ... its all a bit unexplored for me so I cant really add anything concrete - nice job 19:41
weekly: github.com/FCO/ASTQuery/blob/main/REFERENCE.md
notable6 librasteve, Noted! (weekly)
SmokeMachine librasteve: thanks! :) AI helped me with the README as well, if you'd like to give it a read: github.com/FCO/ASTQuery 19:43
ab5tract SmokeMachine: that is genuinely impressive. It’s a bit intense seeing the full volume of RakuAST classes all in one place tho! 19:48
librasteve the readme helps a lot ... I have dabbled in the very shallow end of AST and found it very turgid to deal with ... so I grok that a tool ike this is a big big improvement. would you mind posting an "idiots guide" on www.reddit.com/r/rakulang/ or wherever... if you can think of "the code box to do on raku.org but with extended explanation maybe? 19:50
SmokeMachine librasteve: I'll try to ask opencode to help me preparing something for that... 19:53
librasteve =b 19:54
SmokeMachine here is an example of usage: github.com/FCO/Acme-Overreact/blob...ct.rakumod 19:55
librasteve as a certified idiot, sadly that goes over my head 19:59
SmokeMachine ab5tract: I agree... but I couldn't find a better way to show that... :(
librasteve_: usercontent.irccloud-cdn.com/file/.../image.png 20:00
librasteve not to worry - RakuAST is a big step change and we will work out best way to grip it and to show it off 20:01
jubilatious1_98524 m: my $poet1 = 'Blake'; my $bl = '', my $nil = Nil; say $poet1 // $poet1.so ?? $poet1 !! 'no return' ; 20:25
evalable6 Blake
Raku eval Blake
jubilatious1_98524 m: my $poet1 = 'Blake'; my $bl = '', my $nil = Nil; say $bl // $bl.so ?? $bl !! 'no return' ; 20:26
evalable6 no return
Raku eval no return
jubilatious1_98524 m: my $poet1 = 'Blake'; my $bl = '', my $nil = Nil; say $nil // $nil.so ?? $nil !! 'no return' ;
Raku eval no return
evalable6 no return
jubilatious1_98524 (I screwed up the semicolons between my declarations, but the code still worked) 20:29
20:50 zetaaaa joined
ab5tract SmokeMachine: I didn’t mean it as a knock on the current presentation. It looks great to me. Just a bit fazed by the sheer scope of RakuAST 21:08
21:12 jmcgnh left 21:14 manu_ is now known as eseyman 21:46 TempIRCLogger left, TempIRCLogger joined 21:47 TempIRCLogger left
SmokeMachine I'm trying to write the post about ASTQuery... It seems most of rakuast attributes are RO... that limits a lot the possibilities... 21:48
21:48 TempIRCLogger joined 21:49 jmcgnh joined
lizmat SmokeMachine: a RakuAST is supposed to be immutable from the outside 21:49
SmokeMachine lizmat: even on CHECK?
lizmat well... that's the question, really :-)
do we want to make changes to the tree itself, or create a copy with changes 21:50
SmokeMachine but, I have found a few things mutable before, for example the args of a function... 21:51
I tried the making the copy to... but it seems $*CU is also RO...
lizmat sure, there are some mutator methods... but they are intended to be used by the grammar to fix chicken/egg problems
SmokeMachine I think RakuAST will be much more powerful if editable on CHECK... 21:53
lizmat agree that the RakUAST tree will need to be mangled for all types of uses: think static opts, constant folding, etc... 21:54
but whether that should be attainable by modifying an existing tree, I'm not sure 21:55
SmokeMachine I don't see a huge gain to make all nodes imutable if you can change the root... but I may not be seeing something... 21:56
lizmat consider that a lot of RakuAST nodes contain internal state
21:57 bolangi joined
SmokeMachine that will be needed on the copy as well, right? 21:57
lizmat but then you would only need to write logic that creates internal state on instantiation of a node 21:58
otherwise you would also need to write logic for the internal state on alteration of a node 21:59
whether immutable or mutable is the best way to go, is *not* that clear to me at the moment 22:00
but maybe I'm more functional programmer than the pope here :-)
SmokeMachine lizmat: I had not thought about the state creation... that makes sense... 22:07
I think I'll change ast-query to return a copy of the tree then... maybe create a new func for that... 22:13
22:41 zetaaaa left, zetaaaa joined 23:26 librasteve_ left 23:46 cryosis left, cryosis joined 23:47 cryosis left, cryosis joined 23:51 cryosis left 23:52 cryosis joined, cryosis left 23:53 cryosis 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!