Hello.
Since codec instantiation happens like this:
we have e.g. (new Tapir{}.emptyIO == new Tapir{}.emptyIO) == false
This is obviously a synthetic example but a more general question is IMO valid: given that EndpointIOs are case classes, EndpointIO.Info is a case class, Schema is a case class, CodecFormats are case classes, do you think an e.g. Codec.id (which is a product of a Schema and a CodecFormat) could be structurally comparable as well?
Since encoding/decoding functions are involved, any equality would be accidental anyway. That is, you can always create an “identical” codec but with a different instance of the e.g. id function. So this might be misleading. Even in the schema, there are functions: validators, coproduct projections etc.
Ok, but would you say 2 emptyInputs should be equal to each other?
I don’t think they should be comparable, so equality wouldn’t make sense for them.
What would you say is the right way to check for empty input then? If there’s a better way then == which I used until now, I’ll just do that
Maybe you can share your use-case?
The simplest way would be to check if the input is of type EndpointIO.Empty
My use-case was basically doing some endpoint analysis where i was filtering with sth along the lines of
.filterNot(_.securityInput == emptyInput)
which unexpectedly (for me) failed when i started declaring endpoints in a class that did extends Tapir (regardless of how smart is that, i mean… it’s public so why not :D)
So basically was just looking to get only endpoints that were declaring securityInput and stumbled over universal equality.
... extends Tapir is definitely valid, also part of the docs 
But for your use-case, I guess the isInstanceOf-check for EndpointIO.Empty should work just tine.
Note in case anybody else comes by looking… I found “part of the docs” link broken, but believe I found where it moved to is: part of the docs