General statechart chat (https://statecharts.dev / https://github.com/statecharts/statecharts/discussions) — for XState Q&A → https://github.com/davidkpiano/xstate/discussions
In v4, you can use
pure() to ensure correct action orderAnd v5 already supports the correct order of actions with
assign; you can NPM install the next branch from GitHub locally: https://remarkablemark.org/blog/2016/09/19/npm-install-from-github/
Next after actions are not been assigned to actions
5 replies
Hi,
Playing with XState actions+guards.....
From my experiments today, an action wont get executed if its guard evaluates to false
The docs don't refer to such use-case
Anyone experienced the same? or different?
"processResults": {
"on": {
"Result": [
{
"internal": true,
"actions": "updateCounter",
"cond": {
"type": "someGuard"
}
}
]
}
}"processResults": {
"on": {
"Result": {
"internal": true,
"actions": [
"updateCounter",
{
"type": "xstate.send",
"event": {
"type": "checkCounter"
}
}
]
},
"checkCounter": [
{
"target": "success",
"cond": {
"type": "someGuard <=== This transition will get selected if this condition passes (after the updateCounter took place)
}
},
{
"internal": true <=== This transition will get selected if the above someGuard will return false.
}
]
}
},
1 reply
i love your xstate library.
one question: i couldn't figure out from the documentation is when i invoke a spawning of a new machine and give it a custom name, how exactly do i stop the machine?
- maybe i missed it, but none of the objects seemed to have a
.stop()method.
another question: i have my state machines running, but sometimes the process crashes or i do not expect a specific network event any time soon (which would trigger the next transition)- so i'd like to either temporary shut down or in case of crash, restart a machine, but it needs to be in a specific state with a specific context.
- i can call
.withContext(...), but probably again a bignner problem - couldn't figure out how to make it start out in one or multiple states (in case its a parallel machine or something)
2 replies
Felt challenged by a tweet from @davidkpiano. https://twitter.com/DavidKPiano/status/1374752085938466816
A circuit breaker machine https://stackblitz.com/edit/circuit-breaker — any feedback will be appreciated.
In case someone is interested, I decoupled first version of Visualiser into npm-package
Is there a standard pattern for actors who have a dependency on data managed by another actor?
Let's say I have a CheckoutMachine and a CartMachine, and the CheckoutMachine needs to create or update a payment intent any time the items change in the cart.
Which of these is more idiomatic?
- Use the data from the broadcasted
CartMachineevent to fetch/update the payment intent - Store the data from the event on context, and then use context data to perform the fetch
1 reply
Hi all, I was impressed once again with the simplicity and yet powerful concepts of statecharts. I was able to really improve my code with the concepts together with the reducer hook.
I was wondering if anyone knows articles/papers comparing statecharts to category theory it seems to me that statescharts and the actor model can be expressed as categories (or not, idk). But if you know any articles related to that please do share. Thanks.
I have an example and can show the issue - here it is
2 replies
- empty
- single
- multiple
to capture the three different ways the list can behave.
hey guys. Been struggling to model a kid's video game with state machines. So... the screen is populated with dots. The user is asked to perform various commands that changes the game state. (Ex: Shake the screen, press a green dot, etc).
For simplicity, there's the (1) "Press a X color dot" command, (2) the "Rotate the screen" command, and (3) the "Shake the screen" command.
The game can go to (1) if a dot of the given color exists and if the screen hasn't been rotated. When the user presses the given colored dot, it will add a random colored dot. (2) occurs as long as the screen isn't already rotated. A rotation undoes a shake. (3) Shaking the screen undoes a rotation. We care about this because the command shown to the user is one of "Shake the screen" or "Shake the screen again".
So part of the problem is that as you can see the commands are tightly coupled. If I add another one, the chance of adding a bug is fairly high. So I'm struggling to model the game state correctly with state machines.
1 reply
assume a statemachine, that switches to a state automatically after a delay. e.g.
`<onentry>
<send event="autoLock" delay="5s" id="idleTimer"/>
</onentry>It is loaded in a program from a scxml file.
How would you handle customizing the delay? So, if a user wants to increase the delay to 10s, would you still use the statemachine delay facility? This would mean, that the statemachine at runtime is different from the statechart. Or would you remove the delay from the statechart (scxml) and just call a timer that has a configurable delaytime?
I am not sure, if it is even possible to change the delay after the statemachine was loaded from the scxml file. That is, using QScxmlStateMachine.
2 replies
