pg query! err

[fanhaining]

when parameters more than 10 I get a err

error[E0271]: type mismatch resolving <sqlx_core::postgres::database::Postgres as sqlx_core::database::Database>::Arguments == (&std::string::String, &std::string::String, &i32, &i32, &i32, &i32, &std::string::String, &std::string::String, &i64, &i32)

expected struct sqlx_core::postgres::arguments::PgArguments, found tuple

[mehcode]

@243011068 Thank you for reporting the issue. We'll look into resolving this, nicely, if possible.

---

@abonander I think we can make this work without the tuple impls. Maybe we should just iterate over the bind arguments and run .bind with the macro.

[abonander]

What I wanted to accomplish with using the tuple is having some sort of typestate that blocks the user from accidentally adding superfluous binds. Right now QueryAs won't let you call .bind() if the arguments type is a tuple.

I do see the limitation here, I'm just wondering how do we prevent the aforementioned or if we care.

[mehcode]

@abonander What about an "anonymous" (scoped) tuple struct that you manually implement IntoArguments for?

[abonander]

That would certainly work, no one's complaining about metadata bloat yet so we can probably get away with it.

[mehcode]

It's still a lot less than if we added x32 tuple impls.

[abonander]

Certainly.

[abonander]

Oh, no it won't work for MySQL because we can't know the types. I guess we could generate a struct where every field is generic over Encode.

[abonander]

Alternatively, what if we used &[&dyn Encode<DB>]?

[abonander]

We decided to build the actual PgArguments or MySqlArguments struct and then wrap it in a newtype that implements IntoArguments, this way the typestate on QueryAs that prevents extra .bind() calls is preserved.