#async #rx #async-io #reactivex #rxrust

another-rxrust

A different implementation than rxRust for easier use of ReactiveX in Rust

36 releases

0.0.46 Mar 26, 2024
0.0.45 Apr 22, 2023
0.0.26 Mar 31, 2023

#168 in Asynchronous

MIT license

220KB
8K SLoC

another-rxrust

why new implementation?

rxRust is a Rust language implementation of ReactiveX, but it is not suitable for complex combinations of observable. For those who use ReactiveX in other languages such as rxjs(TypeScript) or rxcpp, rxRust is a very difficult library.

Therefore, I created another-rxrust, thinking that I needed a library that allows observable to be connected in the same way as other platforms, and that ReactiveX can be enjoyed in Rust.

In addition, ReactiveX may not be the best solution if the purpose is to parallelize heavy processing and speed it up. However, ReactiveX is one answer for complex combinations of non-blocking I/O and error handling.

Implementation policy

  • It is assumed that the values and functions that can be emitted may be shared between threads.
  • Value to emit should be Clone + Send + Sync only.
  • Use move to emit values ​​as much as possible.
  • Functions should be Fn() + Send + Sync only.
  • Errors are type-erased using std::any.
    • Thus requiring a 'static lifetime.
    • You can use anyhow.
  • Prioritize flexibility over memory efficiency and execution speed.

Implementation status

See implementation status.

Installation

cargo add  another-rxrust

or

[dependencies]
another-rxrust = {}

If you want to use it on the web, set it as follows.

[dependencies]
another-rxrust = {features=["web"]}

Samples

from_iter, map, zip

use another_rxrust::prelude::*;
use another_rxrust::{print_next_fmt, print_error, print_complete};

fn main() {
  let ob = observables::from_iter(0..10);
  ob.zip(&[ob.map(|x| x + 10), ob.map(|x| x + 20)]).subscribe(
    print_next_fmt!("{:?}"),
    print_error!(),
    print_complete!(),
  );
}
// [console-results]
// next - [0, 10, 20]
// next - [1, 11, 21]
// next - [2, 12, 22]
// next - [3, 13, 23]
// next - [4, 14, 24]
// next - [5, 15, 25]
// next - [6, 16, 26]
// next - [7, 17, 27]
// next - [8, 18, 28]
// next - [9, 19, 29]
// complete

subject, scheduler, take, sample

use another_rxrust::prelude::*;
use std::{thread, time};
use another_rxrust::{print_next_fmt, print_error, print_complete};

fn main() {
  let sbj = subjects::Subject::new();
  observables::interval(
    time::Duration::from_millis(100),
    schedulers::new_thread_scheduler(),
  )
  .sample(sbj.observable())
  .take(3)
  .subscribe(print_next_fmt!("{}"), print_error!(), print_complete!());

  (0..3).for_each(|_| {
    thread::sleep(time::Duration::from_millis(500));
    sbj.next(());
  });
  sbj.complete();
  thread::sleep(time::Duration::from_millis(500));
}
// [console-results (Depends on execution environment)]
// next - 3
// next - 8
// next - 13
// complete

just, error, emptry, never, flat_map...

use another_rxrust::prelude::*;
use std::{thread, time};

fn main() {
  // observable creator function
  fn ob() -> Observable<'static, i32> {
    Observable::create(|s| {
      s.next(100);
      s.next(200);
      s.complete();
    })
  }

  observables::from_iter(1..10)
    .observe_on(schedulers::new_thread_scheduler())
    .flat_map(|x| match x {
      1 => observables::empty(),
      2 => observables::just(x),
      3 => ob().map(move |y| (y + x)),
      4 => observables::error(RxError::from_error("some error")),
      _ => observables::never(),
    })
    .map(|x| format!("{}", x))
    .on_error_resume_next(|e| {
      ob().map(move |x| {
        format!(
          "resume {:?} {}",
          e.downcast_ref::<&str>(),
          x
        )
      })
    })
    .subscribe(
      |x| println!("{}", x),
      |e| println!("{:?}", e.downcast_ref::<&str>()),
      || println!("complete"),
    );

  thread::sleep(time::Duration::from_millis(500));
}
// [console-results]
// next 2
// next 103
// next 203
// next resume "some error" 100
// next resume "some error" 200
// complete

Dependencies

~0–2.3MB
~43K SLoC