interesting idea, I see this more in asyncio itself than aiohttp.

Nice!

1. I think regular context manager is enough here.
2. While I can see the cases for having this, I think better not since this will make code reading more complicated. You need to be sure, that exact block may run no longer than specified timeout.
3. That would be expected.

However, I have bad feelings that this context manager wouldn't work correctly for the tasks that are spawned from the timeout block:

async def bar():
    await asyncio.sleep(3)
    print('hello!')

async def test():
    async with Timeout(1):
        t = asyncio.ensure_future(bar())
        await asyncio.sleep(1)
        await t
    await asyncio.sleep(3)

asyncio.get_event_loop().run_until_complete(test())

I was expected to see an exception, or Task was destroyed but it is pending!, but not "hello!".
For the case when I do await asyncio.ensure_future(bar()) it indeed doesn't makes bar to be completely executed.

Looks like we should make the same guarantees for Timeout as async_for has.

I ported tests from asyncio.wait_for. And added blocking feature like in wait_for:

async with Timeout(None):
    t = await future

in this case timeout is not calculated, similar to await wait_for(future, timeout=None)

as result of discussion in IRL:

1. timeout=None removed
2. cancel api removed
3. updated tests

raise_error argument removed

This thread has been automatically locked since there has not been
any recent activity after it was closed. Please open a new issue for
related bugs.

If you feel like there's important points made in this discussion,
please include those exceprts into that new issue.