Output validation

[matthiasbayer]

Hey Edoardo, great project! Are there plans to add output validation?

I think it would be very handy and also make the migration from tRPC a bit easier:

// Current

const OutputSchema = z.object({ id: z.string(), foo: z.string(), bar: z.number() })
export const myAction = actionClient
  .schema(z.object({ id: z.string() }))
  .action(async ({ parsedInput: { id } }) => {
    return OutputSchema.parse(await Service.getEntity(id))
  })

// Suggested
export const myAction = actionClient
  .schema(z.object({ id: z.string() }))
  .schemaOutput(z.object({ id: z.string(), foo: z.string(), bar: z.number() }))
  .action(async ({ parsedInput: { id } }) => {
    return await Service.getEntity(id)
  })

Could add aliases (input => schema and output => schemaOutput) to make it a bit more intuitive.

Let me know what you think. Cheers!

[TheEdoRan]

Are there plans to add output validation?

Actually, yes. I've been wanting to implement this for a while now.

Could add aliases (input => schema and output => schemaOutput) to make it a bit more intuitive.

I'm thinking of aliasing inputSchema to schema and add outputSchema. I really want to keep the "schema" word in the method name to avoid confusion about the input/output validation and parsing, which is performed when the action is called, and not in inputSchema/outputSchema.

[TheEdoRan]

Update: support will land in the library when #250 gets merged. Added outputSchema method to allow for optional action's output data validation. schema method is left unchanged to avoid confusion with weird aliases, should be good enough, at least for now.

[TheEdoRan]

Implemented in v7.8.0