Hi @Algot,
I released Octans v0.1.0, I implemented the new algorithm as discussed here: https://tilde.zone/@andinus/105559855006590792
It's a lot faster now & I improved the output to return the solution grid. Also, this was my first upload to CPAN (collection of #Perl + #Raku modules).
And it found 10 solutions to this puzzle: https://mastodon.art/@Algot/105413312119416356 - https://asciinema.org/a/385598
Next step would be to solve each starting position concurrently. I'll implement that some time later along with the CGI script!
@Algot If you want to look at the source then follow this link: https://git.tilde.institute/andinus/octans/tree/lib/Octans/CLI.rakumod
GitHub hasn't added syntax highlighting for `.rakumod' yet. `CLI.rakumod' is where octans enters the script.
The file is small because I moved all functions to their own modules, you can find that in the `lib/' directory itself.
Also an interesting thing about Raku: You can use non-ascii characters as variable names, search for `my %𝒻𝒶𝓃𝒸𝓎-𝒸𝒽𝒶𝓇𝓈' in that file. It's a hash map which uses non-ascii characters as variable name.
This version includes a dictionary with it: https://git.tilde.institute/andinus/octans/tree/resources/mwords
It's taken from the Moby Project: https://en.wikipedia.org/wiki/Moby_Project and contains 354,984 words.
And here is the distribution: https://modules.raku.org/dist/octans:cpan:ANDINUS
That means you can install it with zef (Raku module installer), just run `zef install octans' and start using it!
If you have a Linux/BSD machine nearby then I could help you setup Raku. Latest binaries are available for ubuntu, should be the same for any other popular distro.
@Algot Thanks 
,
> including the ability to solve more generic wordsearch puzzles of many different sizes?
Actually that was possible in the online version I linked but not in the program because there octans had to get the starting points by itself.
But since you mentioned it, I fixed the program. There are 2 ways of giving it a puzzle, one is via a file & other, linking to your posts. Both ways, it can solve puzzles of any input size. (see attached image).
Also, you don't have to input the @gray-squares manually, it'll automatically set it. ( sample input file: https://github.com/andinus/octans/blob/master/resources/sample_input )
> I am, in fact using Kubuntu.
Ah, that's nice. It'll be easy to get Raku setup on there. You can use octans right after you get rakudo setup, even without installing zef & the dependency.
> Hope you are getting enough sleep while you devote so much time to this project.
Hah... I should goto sleep 
. Night o/
--
While testing I found another puzzle that has multiple solutions: https://mastodon.art/@Algot/105333136907848390 (see attached image for solutions, some are repeated because there are multiple paths to that word)
You have given me an important reason to install your program on my own computer.
While I attempt to visualize other words as I do my manual puzzle creation, your multi-answer run makes it abundantly clear I can and do mess up.
I didn't know of the words "chalcus" or "canions" before today, having to look them up! I would never have caught them on my own.
@Algot Those words do look uncommon. I plan to write a word search puzzle generator too... someday.
Do you know about tildeverse.org? tildes are pubnixes, anyone can ask for an account.
You don't have to setup Raku on your system, you could ask for an account on one of the tildes & use it there!
@andinus
Wow!
It looks like you've made some great improvements,
(including the ability to solve more generic wordsearch puzzles of many different sizes?)
I am, in fact using Kubuntu.
I'll look into doing an install of Raku, etc. in the next weeks.
(By then, you might even be at version 3.7... 😀 )
Hope you are getting enough sleep while you devote so much time to this project.
Thank you.