search

bmuller / kademlia

A DHT in Python using asyncio
http://kademlia.readthedocs.org
MIT License
830 stars 210 forks source link

_nodesFound sometimes returns a coroutine, which set_digest tries to use as a list #38

Closed jMyles closed 7 years ago

jMyles commented 7 years ago

I've been hunting the cause of this error:

    self.log.info("setting '%s' on %s" % (dkey.hex(), list(map(str, nodes))))
E   TypeError: 'coroutine' object is not iterable

I think I've found it, but what I'm looking at doesn't make much sense.

set_digest calls await spider.find(), calling the result nodes.

This usually works just fine, because nodes ends up being a list, which is used to emit a log on the following line:

self.log.info("setting '%s' on %s" % (dkey.hex(), list(map(str, nodes))))

However, there is at least one execution path where nodes is a coroutine instead of a list.

set_digest calls NodeSpiderCrawl.find, which calls SpiderCrawl._find, which calls NodeSpiderCrawl._nodesFound. You can see from the conclusion of this method that, if we haven't contacted all of self.nearest, then we return the find() method (a coroutine) instead of the list:

https://github.com/bmuller/kademlia/blob/master/kademlia/crawling.py#L142

bootstrap seems to handle this scenario, but set_digest does not.

Am I doing something wrong?

bmuller commented 7 years ago

Hey @jMyles - I think this was fixed in #34 for the python3.5 branch. It wasn't backported to the Python 2 master branch - mostly cause I think that'll just be deprecated here shortly.

jMyles commented 7 years ago

Ahh, @bmuller indeed you are correct.

We're still left hoping for maintenance of the twisted branch, as we have decided to use twisted in the codebase in question.

Thanks as always! Hope you are well.

bmuller commented 7 years ago

K - I'll try to keep that in mind for future bug fixes.