Feature: `memoize` result of asynchronous schema

[GregOnNet]

Preface - valibot is awesome

Thank you all for developing this library. It is the most type-safe library, I have ever used. It comes with great DX & the docs become better and better.

Use Case

Forms:

• Sometimes we need to validate certain fields asynchronously.
• It would be nice if the schema could cache certain inputs and the results of the asynchronous validations to not trigger asynchronous operations if not needed.

New action memoize

Wouldn't it be nice if we could memoize the executions inside a schema, where it is needed? We could provide a synchronous and an asynchronous version...

memoize
memoizeAsync

Example

v.pipeAsync(
  v.string(),
  v.memoizeAsync(
    v.customAsync(input => fetch(`/api/email/${input}`).then(() => /* ... */)), { 
    buffer: 10, // default 1 - tells how many inputs with their respective result are cached
    itemExpiryInMs: 30_000  // default 5_000
  })
)

[fabian-hiller]

Great idea! What is buffer doing? We could work on memoize while also working on a debounce function.

[GregOnNet]

Oh, yeah buffer… I thought it would be handy to be able to tell how many items are cached...

Example:

1. Someone types gregor@domain.de which triggers an API call from the async validator
2. Then someone corrects this to gregor@domain.uk which is calling the API again
3. Finally, someone switches back to gregor@domain.de which is not calling the API because the outcome of the validation is still cached and not outdated.

I hope clarifies the use-case a bit better.

I also like the idea of having debounce for certain values. 💪🏻

[GregOnNet]

After writing the example above, I started thinking if we are talking about cache & cacheAsync instead of Memoization…

[fabian-hiller]

I posted this issue on X to get more feedback and ideas from other users. I am currently focusing on Valibot v1 and Standard Schema and think I will work on this issue after our stable v1 is out.