Every sqlite connection to in-memory db creates a new instance of DB. So if you use a connection pool you can't e.g. apply your migrations and continue using the same DB which could have been useful for tests.
But suddenly thanks to @remmy there's a workaround with named in-memory nodes!
TLDR your connection string is:

sqlite:file:foo?mode=memory&cache=shared

via https://github.com/launchbadge/sqlx/issues/2510#issuecomment-1750535623

#dev #sqlite