Andrey Rakhmatullin

Did you run that with your changes?

In general, as we run that code with `python3 -m scrapy.utils.benchserver` we don't have these problems, and I'm not sure if name clashes are relevant unless you add their paths...

> python scrapy/utils/benchserver.py successfully runs the bench server, though it raises a non-breaking AttributeError in the terminal. I didn't notice it was non-breaking (but still). > Given that you mentioned...

~~Needs lxml.~~

`download()`, the public one, suggested on https://docs.scrapy.org/en/latest/topics/item-pipeline.html#take-screenshot-of-item

Analysis updated for current master with more deprecation removals

As a hack I was able to make `pytest-asyncio` use the existing loop (one created when the reactor is installed) by defining a `_session_event_loop()` fixture, but the next problem seems...

> Could you say some more about the situation that makes this needed? Mostly just fill in some of the details as I think this through (maybe). I have a...

For the record, while moving stuff to a separate thread was harder than I thought, I was able to omit `pytest-asyncio` and `pytest-aiohttp` altogether in my project, by copying the...

Done, except for `scrapy.logformatter.LogFormatter` (which is possible to refactor but isn't a priority and shouldn't force this ticket to be kept open).