Austin Bonander

I plan on improving the `UNNEST($1, $2, $3)` approach by adding an adapter that lets you bind iterators directly so you don't need extra allocations to do the Array-of-Structs to...

Just realized though that `PgArray` would need `&mut` on the iterator during encoding but `Encode::encode()` takes `&self`. Anyone think `RefCell` is a bad idea here? Or maybe it can just...

@markose yes the query is prepared and cached if you use the `sqlx::query()` interface (both the functions and the macros), but keep in mind that you're going to be generating...

You don't need a `Mutex`, you can do a bulk delete using the same `UNNEST` trick as `DELETE` supports it with `USING ...`, but it's overkill to match on _all_...

My ideal would be a type that implements `Encode` when `T: IntoIterator` and `Decode` when `T: FromIterator` so you can use it in both directions. The current `Encode`/`Decode` trait story...

@jyn514 can you provide a `explain verbose` for that `INSERT INTO ... RETURNING release_id` query? A minimal reproduction with schema would be helpful for testing. We can feasibly detect simple...

I'd like to see the workflows running, not sure why they're not.

PRs from first-time contributors require approval every time a workflow wants to be run. It's a bit annoying but it seems to be a decent precaution as an attacker could...

Errors are not typically `Clone`, though. Just off the top of my head, none of these errors from popular libraries are `Clone`: * `std::io::Error` * `hyper::Error` * `reqwest::Error` * `actix_web::error::Error`...

If it's as easy as you say it is, why not open a PR?