Pulsar: Concurrent framework for Python • Hacker News Board

Re: Pulsar: Concurrent framework for Python

#11

Post by hymloth » Mon, Oct 14, 2013, 9:58 AM UTC

gcb0 wrote: post #8

hymloth wrote: post #2

Interesting! Seems like a hyper-framework, a one-fits-all solution for a python backend. I would like to see any benchmarks comparing this to tornado/twisted/nodejs and perhaps an async mongoDB client implementation. Great work anyway!

Thinking about new tech in term of performance is silly. If you need performance NOW just stick with the known solution because odds are you also need stability and a well tested platform.

1) Are known solutions always performant (which is a broad topic)?

2) "I would like to see any benchmarks comparing this to..." != "I only care about performance and nothing else matters"

Re: Pulsar: Concurrent framework for Python

#12

Post by lsbardel » Mon, Oct 14, 2013, 10:51 AM UTC

drdaeman wrote: post #9

Semiofftopic: Animated favicons are evil. Can't just leave the tab open to take a look at a later time.

Fair enough, I've replaced it with a static icon.

http://pythonhosted.org/pulsar/index.html

Re: Pulsar: Concurrent framework for Python

#13

Post by kingkilr » Mon, Oct 14, 2013, 12:20 PM UTC

joshz wrote: post #7

RussianCow wrote: post #5

Finally a concurrency framework that works with Python 3! I will definitely have to check this out, because Twisted is the only thing preventing me from switching (and Gevent doesn't support it either).

What about concurrent.futures from standard library?

It's not really a concurrency framework, just some tooling. It does absolutely nothing to provide evented-IO for example.

Re: Pulsar: Concurrent framework for Python

#14

Post by lvh » Thu, Oct 17, 2013, 9:20 AM UTC

RussianCow wrote: post #5

Finally a concurrency framework that works with Python 3! I will definitely have to check this out, because Twisted is the only thing preventing me from switching (and Gevent doesn't support it either).

The porting process is slow but sure, fortunately. Some parts already work. It'd be cool if we stopped hitting broken stdlib functions. (Turns out that sometimes regular quotes did actually mean bytestrings.)