[00:38] *** deoac left [01:15] *** thowe left [01:15] *** thowe joined [06:11] *** jgaz left [06:13] *** jgaz joined [07:41] *** Manifest0 joined [08:06] *** dakkar joined [08:21] *** tea3po left [08:21] *** tea3po joined [10:45] *** guifa left [14:56] *** SmokeMachine left [14:57] *** mjgardner left [14:57] *** tbrowder__ left [14:59] *** mjgardner joined [14:59] *** tbrowder__ joined [14:59] *** SmokeMachine joined [15:49] Hello Anton, I started using it. There is an issue with people that do not have enough "quota" - will put some money up. The error message is fine, however, for people new to Raku the "Use of uninitialized value element of type Any in string context. ... " is scary. error => {code => insufficient_quota, message => You exceeded your current quota, please check your plan and billing details., param => (Any), [15:49] type => insufficient_quota} Use of uninitialized value element of type Any in string context. Methods .^name, .raku, .gist, or .say can be used to stringify it to something meaningful. in method eval at /usr/local/Cellar/rakudo-star/2023.08/share/perl6/site/sources/2CBCBF8A4A1670092BE1E1523080764269775230 (LLM::Functions::Chat) line 80 in method eval at [15:49] /usr/local/Cellar/rakudo-star/2023.08/share/perl6/site/sources/2CBCBF8A4A1670092BE1E1523080764269775230 (LLM::Functions::Chat) line 56 in block at line 1 in method eval at /usr/local/Cellar/rakudo-star/2023.08/share/perl6/site/sources/C24820763DE9E6A4B431D7F0A8945616880E1EA8 (Jupyter::Kernel::Sandbox) line 139 in code at [15:49] /usr/local/Cellar/rakudo-star/2023.08/share/perl6/site/sources/DA76AEE5D68CD02B4E42BCB2570746B82B483811 (Jupyter::Kernel) line 147 [15:50] I will record issues in github [15:55] @rcmlz Thanks a lot! At this point I am really not sure what is the best way to handle these kind errors. This gets more complicated because different LLMs handle these kind of errors differently. [15:57] PS: how far do I get for 10$? [15:59] Hmm... Which LLM service you are using? I assume one of of OpenAIs : but which one? (The "regular", or the "ChatGPT plus".) [16:01] Another remark on error handling: PaLM often produces moderation errors. It is a problem with PaLM (which I have reported), but also it is better if there as uniform way of handling moderation errors that is also adopted by the different concrete packages for LLM access. [16:06] @rcmlz My bills with OpenAI's regular service rarely exceed $10 per month. I also I have "Plus" which is with a flat rate of $20 per month. (If it is not clear, one pays for these services separately.) [16:45] *** dakkar left [17:44] Hello again! Just a thought that crossed my mind: Suppose I have a large string in which I’d like to make several substitutions. The substrings I wish to substitue might (or might not) have something in common, e.g. they might all start with the same character. Should I do well to perform several substitutions on that string in order, or is there a more efficient, but comparably simple way, where Raku could perform all of the substitution [17:44] s ‘in one fell swoop’, so to speak? [17:47] I imagine a custom procedure could be written (and perhaps generalised), but I wonder if there is some straightforward method or ‘good practice’ for this kind of thing. cx [18:21] Tirifto_: You can use .subst with a Callable to that. But that requires a matcher that can match all needles in one go. [18:25] Tirifto_: my approach would be to do them sequentially… with normal sized strings this should fit in RAM so I don’t see much performance gain for parallelism and I do see confusion such as what if an early subst contains a later match [18:25] Or .rindex with a list of needles and .substr if you want to make changes in place. You have to work from the end of the string because substitions may otherwise shift indices. [18:27] No wait, that is .indices nowadays. [18:28] Depending on what you want to do, breaking the Str up into a List of Str might be simpler and faster. [18:34] gfldex: does rindex mean reverse index? [18:35] yes [18:35] But it doesn't take a list of needles. [18:36] aha - tx! [18:37] ps. i also would consider “sharding” a very large Str (works of Shakespeare ish) … but dealing with boundaries is then a headache. [18:46] Thank you for the advice, gfldex, @librasteve. I suppose writing a Callable would be fairly straightforward and universal, and so applicable to my use case. But I suppose I’ll only experiment with that if performance turns out to be a problem, which I suppose it won’t. xP [18:54] m: my $s = ‘this is a string.’; my %replace = i => 1, r => 2; say join ‘’, gather for $s.split(, :p) -> $p { $p ~~ Pair ?? take %replace{$p.value} !! take $p } [18:54] rakudo-moar cfcf9fb3e: OUTPUT: «th1s 1s a st21ng.␤» [18:55] This is the most flexible because you can use loop-control to bail early, return Empty and such. [18:56] Also, this shows that gradual typing is just better. :-> [20:21] *** tea3po left [20:21] *** tea3po joined [20:23] *** teatwo joined [20:26] *** tea3po left