Thoughts on resource oriented computing (ROC)

[Kyle Wilt]

I have been a "secret admirer" of clojure and the clojure approach to problem solving for quite some time now even though I'm not really a direct practitioner. I do try to convert my C# code into as "clojure like" a model as reasonably possible given all of my constraints. I have been a big fan of the design approach behind core.async and channels in general.

Recently I came across an approach to building systems that I am very curious what the clojure community would make of. There's a company called 1060research that has been using what they call "Resource Oriented Computing" for over a decade now I believe.  One of its goals is to bring the economics of the model of the web into the level of software components. Their implementation of this approach is called NetKernel which as far as I can tell uses typical Java OO at its core but that implementation detail doesn't completely directly pervade the model it's trying to provide. It does create limitations for those of us who don't use the JVM however. They actually have a clojure language module to support running clojure code in their definition of components.

Here are some links for anyone who might be interested in starting to dig into it:

http://resources.1060research.com/docs/ROCForDevelopers.pdf

You tube video about the high level concepts

So to repeat the purpose of my post here, I'm really interested in how the community perceives this concept of "resource oriented computing" and how it meshes with the clojure mindset to design of systems. From my perspective it doesn't directly clash and in some ways is very complimentary.

I apologize of this topic is inappropriate to this group, I've never posted to any clojure related groups before.

[Tom Hicks]

Hi Kyle,

My memory is that Peter and Tony started 1060 Research almost 20 years ago. They used to publish a fairly frequent email newsletter (http://wiki.netkernel.org/wink/wiki/NetKernel/News/) about their activities but I haven't seen a newsletter from them in over a year and a half.

I mentioned NetKernel to Rich Hickey several years ago and found that he was familiar with it, at the time.

re Clojure: I would caution you that I wrote that initial, limited Clojure module a *long* time ago (2009, I think). It was not multi-tenant capable and should have been rewritten to keep up with the evolution of Clojure. You should check with 1060 Research directly to find out what the current status of the module is.

   regards,

       -t

[Kyle Wilt]

Thanks for the response. What is your opinion of it/experience with it? I'm really wondering how it meshes (if at all) with the way in which people think about software design/construction in the clojure "world".

[Tom Hicks]

I view ROC (as defined by 1060Research and instantiated in NetKernel) as a systems architecture which attempts to bring some of the RESTful principles of the Web into application and systems programming. Every architecture has its points of "fit" which make it good for some problems and not so good for others. A direct comparison of architecture and language is difficult and the Clojure language has proven itself applicable across a wide variety of architectures. While Clojure and NetKernel share some underlying philosophies of software construction, I don't think that makes Clojure significantly "better" than other languages for implementing ROC (and NetKernel was designed to be polyglot, at least across JVM languages).

I would be amenable to more discussion about my experiences with NetKernel offline, but I don't think this really belongs in the Clojure forum.

    regards,

       -t