Fork me on GitHub

I believe @lucasbradstreet discovered this, bug fix is pending in It's tough to reproduce since its non-deterministic when it happens, I think the fix isn't too bad though


Would you guys say that onyx is a solution to the problem of managing dependencies among asynchronous operations?


Yes, I'd say it's an excellent solution for that, especially if you want to ever want to connect up channels between networked nodes (which is something that onyx essentially gives you for free)


We've had users do benchmarks on local nodes and we're within 10% of pure core.async when running locally


cool thanks, I'll make a note of it on SO. I wasn't sure if I needed everything that Onyx brings in (cluster management and all) but after thinking about it some more, starting with Onyx now just gives me more ability to scale up in the future if need be


Yep, that's something we've focussed on a lot. Making it easy to get up and running with one node but able to scale later


Working on pure channels does give you some features like working directly with transducers, so it's not an exact mapping


Onyx 0.8.0 is officially out! Thanks to everyone who helped us get it ready!


Any plans for a book / videos tutorials guys?


@adebesin: Video series already in the works. simple_smile


awesome 👍


learn-onyx is a good way to go through learning the bits and pieces


@lucasbradstreet: Thanks! I have just started playing around with Onyx but still havent figured out what its strengths are. At work we are currently using Spark after moving away from Hadoop and Hive. How does Onyx differ from Spark in the context of functionality ?


Spark is good - I’d say that our main strength against Spark is that we’re more stream orientated, which results in lower latency processing than Spark’s micro batching can


I hate to be cliche, but Onyx's API is markedly more decomplected. There, I used the D word.


And yes, there’s also the usability aspect


great job! 👍


@spangler: How often are you hitting that exception? I think it's exacerbated by that fact that your messages are large and take a while to ack.


@michaeldrogalis About 1 out of 20 jobs... each each job runs hundreds of tasks


Spark as a concept is obviously great but Scala is such a dreadful language far to complex to be any where near productive. I think the Clojure community needs Onyx and we should all get behind promoting its potential. I'm glad you guys took on the challenge.. great work 👍


@spangler: Okay. Sorry, you've admittedly been hitting a lot more bugs than other people because you're using features that have gotten less mileage. This one is particularly annoying because its non-deterministic. I have an idea for a temporary fix to get you going if I cant get a proper fix soon.


@adebesin: Thanks! We really appreciate that! simple_smile


@adebesin It does have great potential


Might just implement that method its complaining about missing on the Function record and no-op it


Build is running, I added this patch: Should pass the tests, hopefully it's enough to keep moving.


@spangler: It passed the tests. Give the latest SNAPSHOT a try. Hope it does the trick for now.