Redo

An undo/redo library.

About

It uses the Command Pattern where the user implements the RedoCmd trait for a command.

The RedoStack has two states, clean and dirty. The stack is clean when no more commands can be redone, otherwise it is dirty. The stack will notice when it's state changes to either dirty or clean, and call the user defined methods set in on_clean and on_dirty. This is useful if you want to trigger some event when the state changes, eg. enabling and disabling buttons in an ui.

It also supports merging of commands by implementing the merge method for a command.

Redo vs Undo

redo uses static dispatch instead of dynamic dispatch to store the commands, which means it should be faster than undo. However, this means that you can only store one type of command in a RedoStack at a time. Both supports state handling and command merging but undo will automatically merge commands with the same id, while in redo you need to implement the merge method yourself.

I recommend using undo by default and to use redo when performance is important. They have similar API, so it should be easy to switch between them if necessary.

Examples

[dependencies]
redo = "0.2.0"

use redo::{self, RedoCmd, RedoStack};

#[derive(Clone, Copy)]
struct PopCmd {
    vec: *mut Vec<i32>,
    e: Option<i32>,
}

impl RedoCmd for PopCmd {
    type Err = ();

    fn redo(&mut self) -> redo::Result<()> {
        self.e = unsafe {
            let ref mut vec = *self.vec;
            vec.pop()
        };
        Ok(())
    }

    fn undo(&mut self) -> redo::Result<()> {
        unsafe {
            let ref mut vec = *self.vec;
            let e = self.e.ok_or(())?;
            vec.push(e);
        }
        Ok(())
    }
}

fn foo() -> redo::Result<()> {
    let mut vec = vec![1, 2, 3];
    let mut stack = RedoStack::new();
    let cmd = PopCmd { vec: &mut vec, e: None };

    stack.push(cmd)?;
    stack.push(cmd)?;
    stack.push(cmd)?;

    assert!(vec.is_empty());

    stack.undo()?;
    stack.undo()?;
    stack.undo()?;

    assert_eq!(vec.len(), 3);
    Ok(())
}

An unsafe implementation of redo and undo is used in examples since it is less verbose and makes the examples easier to follow.