If the GIL is a problem with both approaches, I think the best course of action
would be you just stick with what is already in the Multi-Lang protocol, rather
than adding another thing that Storn libraries will need to support.
Also, as long as the amount of time that a ShellBolt will wait to
Hi hermanth,
The Python multi-lang implementation provided with Storm is not really
production ready for exactly the reasons you outline. At Parse.ly, we use
streamparse for our Python Bolts/Spouts. It has built in logging support and
also takes over stdin/stdout to prevent you from inadvertent
Hi Jungtaek,
Sorry I didn’t notice this earlier, as I was the person who filed STORM–513 in
the first place.
Having just implemented the new heartbeat protocol in Python (for streamparse)
and Perl (for IO::Storm), I’m not crazy about needing to add another heartbeat
approach to multiple librar
As part of the development of streamparse, we have a BatchingBolt that
processes tuples in batches. It's intended for use with things like databases
that are more performant when you send things in batches.
I've recently proposed switching our BatchingBolt implementation over from
using a time