Should we avoid building the queries here if you're going to compile them later?

Hmm, it's been a long time since I wrote this PR, but I think we still need to convert the query to ClauseElement and bind params, but maybe we don't need list comprehension here, so maybe something like that is sufficient

async def execute_many(
        self, query: typing.Union[ClauseElement, str], values: list
    ) -> None:
        query = self._build_query(query,values[0])
        async with self._query_lock:
            await self._connection.execute_many(query, values)

Hm interesting. I'm not sure myself, just wanted to check if you had considered it.

Thanks for chiming in on this after the delay. If you're interested in working on it again I can review and release it. Did you confirm this works and improves performance as expected?

I can work on it tmr for a few.
it should improve the performance but i didn't make any benchmarks tbh

I presume this it not a breaking change for users because this is an internal method and values is always provided by us in the public method?

I'm not sure I understand what you mean here, but I think users are only supposed to use execute_many from the core file, not from any of the backends

Yep that's the question — I want to gauge if changing this function signature is a breaking change.

yea I don't think it should break anything