Jacob Barber

Awesome, thanks for looking in to this! I'll poke at it in a bit.

I'll work on the tests, since we've addressed that classloader issue

Ah I see. I'll see if I can do some hacking. 👍

I came across this same thing upon updating, and really I don't think anybody would ever want to be prompted about this. Disabling the prompt by default is probably the...

To add on to this, a `MERGE INTO` query exhibits strange behavior ```scala sql""" MERGE INTO fail(id, time, bool) KEY (id) VALUES ($uuid, $t, $b) """.update.check.unsafePerformIO /* MERGE INTO fail(id,...

Adding this query to the `h2types.scala` test suite also yields failure: ```scala "pass query analysis for unascribed UUID" in { val a = sql"select random_uuid()".query[UUID].analysis.transact(xa).unsafePerformIO a.alignmentErrors must_== Nil } ```...

I've created a branch with failing test cases demonstrating the issues above: ~~https://github.com/Jacoby6000/doobie/tree/h2-weirdness~~

I updated the branch with a different UUID meta type. I've defined it like: ```scala implicit val UuidType = Meta.advanced[UUID]( NonEmptyListOf(jdbctype.Binary), NonEmptyListOf("uuid", "UUID"), _.getObject(_) match { case null => null...

From what I've gathered the analyzer thinks all types in updates are varchar. Queries do fine for whatever reason. Invoking JDBC directly gives the correct metadata, so I'm not sure...

I've gone ahead and added tests demonstrating this problem in a branch (deleted my old fork, so that old demonstration was dead. This is a better set up anyway): https://github.com/Jacoby6000/doobie/blob/bug/h2-updates/modules/h2/src/test/scala/doobie/h2/h2types.scala