[Autocomplete] Provide more details in the onChange event

[whuey1992]

Is there a way to get only the option that was added/removed in the onChange event handler? Controlled auto complete shows this as their onChange handler:

onChange={(event, newValue) => {
          setValue(newValue);
        }}

So when i use this for multiple my previous state is [A,B] and newValue is [A,B,C]. I have to diff the previous state with next state to see what was added or removed. When I tried using event.target.value, the first onChange returns 0 then after gives me the correct value that changed.

[oliviertassinari]

@whuey1992 What's your use case?

[whuey1992]

I am using two autocompletes for a form. When certain options are selected in one autocomplete, it requires options from the other autocomplete to be selected. Same with if those options are removed.

[whuey1992]

When a new option is selected I can check the last element added to newValue but for deletion I need to do a diff of previous state vs newValue. Which is also kind of made more difficult to distinguish for clear

[oliviertassinari]

@whuey1992 Thanks for sharing. It's unclear what a good solution to the problem would be. I would propose that we wait for more feedback before taking action.

In your case, I don't understand what prevents you to use the final state to run the logic.

[whuey1992]

It doesnt prevent me from running the logic, but it adds complexity to do a workaround (possibility of bugs) and isn't optimal O(AB) to do diffing.

[oliviertassinari]

@whuey1992 Ok thanks for the feedback. I'm closing. From what you are describing, you are trying to bridge a gab between a declarative world and an imperative one. I would suggest that you check how you could remove this imperative requirement.

---

I have added the waiting for users upvotes tag. I'm closing the issue as we are not sure people are looking for such abstraction. So please upvote this issue if you are. We will prioritize our effort based on the number of upvotes.

[oliviertassinari]

On a side note, the fix on our side would be easy. So it really boils down to encouraging stateless logic.