Library interface: Unexpected result when calling run_query directly on a Machine struct

[triska]

Today the following question was raised on the #scryer IRC channel, I file it here as an issue so that we remember it:

Given the fact colleague(joe, mike)., when we query it with colleague(joe, X)., then we get as QueryResolution: Matches([QueryMatch { bindings: {"X": String("mike")} }]).

Question: Why is X is bound to String("mike"), instead of an atom?

Can anyone confirm and explain this, and is more information needed here (@shawa)? Thank you a lot!

[bakaq]

What version is this? This is a problem that was solved with #2475. On current master this works as intended:

use scryer_prolog::Machine;

pub fn main() -> () {
    let mut machine = Machine::new_lib();

    machine.load_module_string("facts", "colleague(joe, mike).".into());
    let result = machine.run_query("colleague(joe, X).".into()).unwrap();

    println!("{result:?}");
    // Matches([QueryMatch { bindings: {"X": Atom("mike")} }])
}

[triska]

Excellent, thank you a lot! I am closing this as resolved!

[shawa]

Looks like this solves the issue; working correctly my end - thanks for taking a look at it!

I'm very new to the project, and Rust, but it looks like this was an issue specifically with the parser - I mentioned on IRC that I'm interested in interop with other languages, and I'm looking at using Elixir as a frontend. It's early days, but I would be even more interested in constructing/manipulating AST directly (bypassing any need to ever actually build/parse Prolog programs/queries as strings altogether), and having a Machine load that.

Is this an avenue the internals are currently amenable to - and if not, where would be a good place to start looking, if I wanted to contribute this capability?

Thanks again for the help!

[triska]

@shawa: There is currently ongoing work to make Scryer Prolog useable from other languages, please have a look: #2465