00:21 dgl is now known as dg 00:43 kjp left 00:44 kjp joined 00:47 kjp left, kjp joined 00:48 [Coke]_ left 01:05 hulk joined, kylese left 02:15 hulk left, kylese joined 02:33 sena_kun joined 02:56 sorenson left 03:35 Aedil joined 04:20 ilogger2 left, ilogger2_ joined 05:30 Sgeo left 06:16 sorenson joined
ab5tract ingy: is that script visible online somewhere? I’m curious what non-trivial yaml script looks like 06:35
07:00 andinus left 07:02 andinus joined 07:38 sena_kun left
ab5tract antononcube: congrats on the new module. I'll definitely give this a try sometime. Is there any way to extend it to storing "analyzed code output" in the vector db? 08:16
08:20 dakkar joined
antononcube Right, that is the “harder” use case. And it is an interesting question: should a certain code base be used “as is”, or be (LLM-)preprocessed to have comments. 08:26
An article can be naturally split into paragraphs and then a vector embedding for each paragraph be obtained. With code the splitting is not that easy or natural. So, another interesting, related question is would partial code fragments be “reasonably” retrieved when using the embeddings. 08:32
Obviously, Raku parsers/grammars or RakuAST can help splitting Raku code into reasonable chunks. 08:42
ab5tract it's still fuzzy to me, but its coming into clearer focus 08:45
09:15 wayland76 joined 10:09 ecocode left 10:14 ecocode joined 11:06 wayland76 left 12:47 sena_kun joined 12:54 [Coke]_ joined
[Coke] rak: Failed to open file /Volumes/deloitte/doc/doc/Type/Range.rakudoc: Too many open files 13:45
lizmat: gist.github.com/coke/913c4279578f9...c8f55e4c68
lizmat what am I looking at there ? 13:46
[Coke] I just ran rak and got that error. 14:17
I can open a ticket and throw in version info if that'll help
ah. it's when I add $(git ls-files) on the command line 14:18
I do that in shell often if I just want version controlled files. looks like rak is performant enough without it, but if I add it, it complains 14:19
lizmat h.mmmm anyways, there's a #raku-rak channel if you have specific rak issues :-)
timo [Coke]: what's your ulimit -n, aka "max open files"? it tends to be rather quite a bit low in normal systems nowadays 14:31
i'm surprised to find mine is 1024 and i've not had anything explode yet
[Coke] 256!
timo holy shit
ab5tract What is this, 1992? :) 14:32
[Coke] yup, upping that to 1024 avoids the explosion.
m: Date.new.year.say
camelia Cannot call Date.new with no parameters
in block <unit> at <tmp> line 1
[Coke] m: Date.today.year.say
camelia 2024
[Coke] nope!
timo m: say [*] 1..256 14:33
camelia 85781777534284265411908227168123262515778152027948561985965565037726945255314758937744029136045140845037588534233658430615719683469369647532228928849742602567963733256336878644267520762679456018796886797152114330770207752664645146470918732610083287632…
timo that's a lot of open files you can have! 14:34
[Coke] irssi question - when I do "/save" or "/layout save", my irssi config is not updated to reflect the current set of open windows. am I missing something? 14:36
timo is it in a secondary or tertiary config file? 14:37
i haven't used irssi in a decade, i've switched to weechat some time ago
[Coke] finds www.linode.com/docs/guides/advance...save-state
add to add a script in an alias. whee. 14:38
*had*
15:36 eseyman_ joined, Voldenet_ joined, tejr_ joined 15:41 eseyman left, Voldenet left, BooK left, Altreus left, tib left, tejr left, destroycomputers left, rjbs left, ky left, timo left, tejr_ is now known as tejr, Voldenet_ is now known as Voldenet 15:43 BooK joined 15:52 nine_ left, nine joined 16:39 dakkar left
lucs Am I misunderstanding the string quoting :f adverb? : 16:56
m: sub baz { $^a }; print Q:f'&baz(123)<foo>'; # Expecting "123<foo>" 16:57
camelia Type Int does not support associative indexing.
in block <unit> at <tmp> line 1
lucs (hungry; back later) 17:01
timo i think the f, c, s, a, h adverbs are about what can start an interpolation, so as long as there are circumfixes later, it will chain? 17:15
antononcube weekly: youtu.be/JHO2Wk1b-Og 17:23
notable6__ antononcube, Noted! (weekly)
17:38 jjido joined
librasteve lucs: looks like you got 123<foo> except that 123 as an Int doesn’t take an associative index <foo> … maybe use parens to disambiguate further? 18:11
tellable6__ 2024-09-16T18:53:45Z #raku <ab5tract> librasteve: syntax highlighting should be fixed :)
librasteve .oO 18:12
scullucs @librasteve : not sure parens where... 18:14
librasteve m: sub baz { $^a }; print Q:f'(&baz(123))<foo>'; 18:18
evalable6 (123)<foo>
Raku eval (123)<foo>
librasteve hmmm
m: sub baz { $^a }; print Q:f'&baz(123) <foo>'; 18:20
evalable6 123 <foo>
Raku eval 123 <foo>
librasteve with a space
looks like we need a non space space 18:21
at this point I would say use Q with curlies - ymmv
scullucs Yeah, turns out curlies would be okay in what I'm actually doing, but I've been using the :f adverb successfully in with generated LaTeX code so much that the "Q:f:..." idiom has become my go to "interpolate only these things, absolutely nothing else" mechanism. 18:25
And so I was wondering what was up with that, and the "chain circumfixes" thing makes sense. 18:26
librasteve yeah - quoting is a gnarly thing ... 18:36
m: sub baz { $^a }; sub nss {"\c[0020]"}; print Q:f'&baz(123)&nss<foo>';
Raku eval Exit code: 1 Type Sub does not support associative indexing. in block <unit> at main.raku line 1
evalable6 (exit code 1) Type Sub does not support associative indexing.
in block <unit> at /tmp/fFh76Sdeob line 1
ab5tract I suspect the issue here is that `:f` specifies "function" calls, which somehow `.AT-KEY` (via `<>`) is being considered under 18:45
huh, but that's not actually true.. 18:46
m: my %a = "a" => "Z"; say Q:f[ $a<a> ];
camelia $a<a>
ab5tract I think this is definitely worth reporting a GitHub issue over 18:47
(also broken in RakuAST) 18:49
timo you can switch to a different quoting mode inside of a qq with \q 19:01
19:05 jjido left
ab5tract that's a useful feature, but is it usable here? 19:14
the RakuAST ouput shows that a `Postcircumfix::LiteralHashIndex` is generated for `<foo>` 19:15
that doesn't seem to be expected behavior for a quoting mode that is supposed to restrict itself to function calls 19:16
timo m: sub lol { "^$_^" }; say q'not: &lol(123) \q:f[&lol(321)]<543> :)' 19:19
camelia ===SORRY!=== Error while compiling <tmp>
Confused
at <tmp>:1
------> sub lol { "^$_^" }; say q'not: &lol(123)⏏ \q:f[&lol(321)]<543> :)'
expecting any of:
infix
infix stopper
postfix
st…
timo m: sub lol { "^$_^" }; say q'not: &lol(123) \qq:f[&lol(321)]<543> :)' 19:20
camelia ===SORRY!=== Error while compiling <tmp>
Confused
at <tmp>:1
------> sub lol { "^$_^" }; say q'not: &lol(123)⏏ \qq:f[&lol(321)]<543> :)'
expecting any of:
infix
infix stopper
postfix
s…
timo m: sub lol { "^$_^" }; say q'not: &lol \qq:f[&lol(321)]<543> :)'
camelia ===SORRY!=== Error while compiling <tmp>
Confused
at <tmp>:1
------> sub lol { "^$_^" }; say q'not: &lol⏏ \qq:f[&lol(321)]<543> :)'
expecting any of:
infix
infix stopper
postfix
statem…
timo not sure what i am doing wrong here tbh?
m: sub lol { "^$_^" }; say q'not: &lol :)'
camelia ===SORRY!=== Error while compiling <tmp>
Unexpected closing bracket
at <tmp>:1
------> sub lol { "^$_^" }; say q'not: &lol :⏏)'
timo oh, ' is allowed in identifiers 19:21
so q' bla ' works but q'bla' doesn't 19:22
m: sub lol { "^" ~ $^a ~ "^" }; say q 'not: &lol() \qq:f [&lol(123)]<321> haha' 19:26
camelia not: &lol() ^123^<321> haha
ab5tract the `<321>` is supposed to be inside of the quote, iiuc 19:34
ah, nevermind, I see what you did there :)
it still feels like a bug to me though
which is to say, this seems inconsistent and I'm not sure how to reconcile the difference: 19:46
m: my %a = :1a; say Q:f[ %a<a> ]; sub h(%h) { %h }; say Q:f[ &h(%a)<a> ]
camelia %a<a>
1
19:51 discord-raku-bot left, discord-raku-bot joined
ab5tract I guess it's this part (from the QAST output): `QAST::Op(call &postcircumfix:<{ }>)` 20:12
librasteve m: sub baz { $^a }; say Q:c|{baz(123)}<foo>|; 20:21
evalable6 123<foo>
Raku eval 123<foo>
librasteve m: sub baz { $^a }; say Q:f|&baz(123)<foo>|;
Raku eval Exit code: 1 Type Int does not support associative indexing. in block <unit> at main.raku line 1
evalable6 (exit code 1) Type Int does not support associative indexing.
in block <unit> at /tmp/Kqyz5d_Iee line 1
librasteve ^^ in my view, these two alternatives should give the same result ... so seems like a bug to me too 20:22
20:25 discord-raku-bot left 20:26 discord-raku-bot joined 20:29 Aedil left 21:09 kylese left, kylese joined 22:03 sena_kun left 23:06 Sgeo joined