Austin Bonander

I'd accept a PR to support this: ```rust #[sqlx::test(tokio(...))] ``` and then we don't have to parse `...`, we can just pass it directly to `#[tokio::test(...)]`. The problem is, that's...

It would be helpful if you provided the actual error message. The only significant change to the internals of the MySQL driver since 0.8.3 would be https://github.com/launchbadge/sqlx/pull/3812

That's likely a result of #3765, specifically this bit of code: https://github.com/launchbadge/sqlx/blob/bab1b022bd56a64f9a08b46b36b97c5cff19d77e/sqlx-mysql/src/transaction.rs#L33-L35 It likely means that StarRocks responded to a `BEGIN` statement with an `OK` packet without the [`SERVER_STATUS_IN_TRANS`](https://dev.mysql.com/doc/dev/mysql-server/9.2.0/mysql__com_8h.html#a1d854e841086925be1883e4d7b4e8cad) flag...

Reported upstream as https://github.com/StarRocks/starrocks/issues/59125

No, but there's also nowhere that we begin a transaction in the code path of opening a connection. The only time the MySQL driver implicitly begins a transaction is when...

https://github.com/ClickHouse/ClickHouse/pull/88818 is released in 25.11 so this can move forward now.

I neglected this for a bit, sorry. I think this is an improvement over the existing forms, we can iterate as things come up.

Yeah, the macros emit instances of `query::Map` that are specifically keyed to the same database they compiled against. I think this is a really important sanity check and I'm loathe...

Alternatively we could support a protocol like `sqlite+anydb://` in database URLs that would cause the macros to emit `DB = Any` instead of `DB = Sqlite` where applicable. I'm not...

[After discussing on Discord](https://discordapp.com/channels/665528275556106240/700206072144855091/796943684985552947), @mehcode and I decided we'll support both options: * `sqlx::macros!(driver = "any")` which can also be used to enforce, e.g. `sqlx::macros!(driver = "postgres")` (will error if...