Alexey Kinev

@jianingy You can try 0.6 pre-release right now, just use `--pre` flag for pip, like that `pip install --pre peewee-async`. Also specifying `peewee-async>=0.6.0a0` should also work I believe. I'm publishing...

Hello @fanjindong! Thank you for reporting and PR, I still don't understand this issue. Guess we need some specific tests here to better understand conditions. From what you're saying I...

We need a policy for such situations. What is expected behaviour? Current implementation assumes that if we failed to acquire cursor, something is wrong with connection and it should be...

Looking at the whole method: ```python async def cursor_async(self): """Acquire async cursor. """ await self.connect_async(loop=self._loop) if self.transaction_depth_async() > 0: conn = self.transaction_conn_async() else: conn = None try: return (await self._async_conn.cursor(conn=conn))...

@fanjindong The difficulty here is to distinguish between two situations: 1. Connection is overloaded by many requests 2. Connection is corrupted and needs to be re-initiated Both scenarios will end...

@fanjindong no problem, it's provided as-is. But I'd keep the issue opened, probably someone will find info here useful.

@kalombos I'm totally for it, will do.

@kalombos sorry for the long response, check the invite

@kalombos I'm also thinking of publishing to PyPi via Github Workflows, should not be a problem

Hi @frosthamster, I'm sure this is due to change of approach that `aiopg` uses to work with loop. I borrowed the explicit loop parameter idea from `aiopg` initially when I...