is there a way to cache the map function of a flow for distinct events so it doesn’t recompute for t...

# getting-started

[allan.conda]

is there a way to cache the map function of a flow for distinct events so it doesn’t recompute for the same value?

flow(1, 1, 2, 2, 3).map { expensiveMapping(it) }.collect()

[itzik kasovitch]

1. I think you don't need the return.
2. I guess you can wrap the implementation of expensiveMapping with something that memoizes the result.

[allan.conda]

updated for 1 🙇

like saving it at a map? That’s our current solution, but I’m not sure it’s thread safe. The flow could be updated from different threads

[itzik kasovitch]

Well, you can use a thread safe map. Are you actually asking if there's a builtin caching mechanism for flow?

[allan.conda]

Yes, basically. Mainly because of this comment when using DataStore on Android

Do not layer a cache on top of this API: it will be be impossible to guarantee consistency.
* Instead, use data.first() to access a single snapshot.

I want to follow it but the map is doing cryptography, and it’s expensive to decrypt the value every time.

[itzik kasovitch]

Got it. I am not sure how to do it.

[allan.conda]

thanks for your insights anyway 🙂

[hfhbd]

Don't you just need to use distinctUntilChanged?