Mike DePalatis

What if you tried something like this as your reload hook: ```python import asyncio from threading import Event def close_connection(): closed = Event() task = asyncio.create_task(aiopg.sa.wait_closed()) task.add_done_callback(lambda: closed.set()) closed.wait() ```

I definitely like the idea of using a `ResultObject` (or maybe just a dict) for the output. In addition to not having to remember the order of positional output, it...

I've started implementing a proof-of-concept results object [here](https://github.com/mivade/allantools/tree/resultobj). It's not ready yet, but I have added an `rtype` keyword argument to calculations allowing easy switching between the current tuple return...

@fmeynadier I'm personally not a fan of that style of giving inputs since it generally seems like it overuses objected oriented features to the point of being too verbose (it...

Looks good. I think you make a good point with using the `compute` method. We might even be able to use some Python magic to automatically map method calls to...

`GistSchema` isn't actually a `Schema` in this case but has an attached attribute to give you the schema class. I've worked around this by just explicitly creating schemas like this...

> Getting PK of inserted rows. Is this currently possible?

Digging into the code, it looks like it is in the same way as with SQLAlchemy (i.e., it gets returned when calling `execute` on an insert statement). Unfortunately this doesn't...

This would be *really* nice to have. Ignoring the `default` value in the table definition currently makes this package not work as an async drop-in replacement for equivalent SQLAlchemy calls....

In the implementation there's no reason for that method to be async. By convention methods that are prepended with an underscore are private and generally shouldn't be overridden by subclasses....