functions
functions copied to clipboard
iron-io
Is this project still active?
I'm really impressed with this project so far and see great applications for this. However given the still early stage of this project and the last commits being months in the past, I was wondering if this project is still being actively developed and maintained?
Hello there, and thanks for checking Iron Functions out !
Recently, iron.io was acquired by Xenon Ventures. Some of the original developers are not with us anymore. So far, we've focused on the main Iron Suite (IronMq and IronWorker) but Iron Functions is going to get attention and an uplift too. So please bear with us until we get up to speed !
Hi there, If you want an active development serverless project, you can use the openFaaS one (https://github.com/alexellis/faas) as an alternative to ironFunction. FaaS is working on multi cloud (public and private) and kubernetes as well. It provides synchronous functions as a service. And asynchronous ones too (where ironFunction failed see my issue).
You can make a 
or start reading the python tutorial here.
IronWorker is perfect for async processing, even for enterprise level projects (CNN/Philips/Veritone)
@rkononov of course ironWorker is working well for async processing, as the process is queued and de-queued when an ironWorker is available and the function is also deferred. But it's bugged when you try to pass data into it. Otherwise, please give me an explanation about my issue. I am sure ironFunction is a enterprise-useful project. It is really simple to use, and your cli allows a lot of good features (building an app from scratch only by managing routes like LEGO® is especially powerful)
In IronWorker you could pass payload using this approach - http://dev.iron.io/worker/reference/payload/ No idea about IronFunctions.
@Tofull We're ramping up development on IronFunctions and you should see a lot of improvements coming soon. We're highly focused on keeping the project open and modular (for example, you may want to collect metrics somewhere else, and not be tied to Prometheus). So, stay tuned!
In regard to deep learning, we recently started supporting GPU instances on AWS via IronWorker (but of course that's our async, non-open source product).
Thanks for checking in!
@dylanz Thanks for the informations. Great blog article ! Keep working hard. I won't be surprised of the ironFunctions' success for Machine Learning community. ;)
@elitan Good Q! We reached a point with the project we're happy with so we ended up stopping active development on it (we have a few production installs that suite us fine at the moment). That said, we're still planning on actively maintaining things and pull requests are definitely welcome.
To the point raised by @Tofull OpenFaaS also supports asynchronous workloads. See the hands-on labs for more: https://github.com/openfaas/workshop
