Gpars dataflow and Reactive streams

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Gpars dataflow and Reactive streams

Paolo Di Tommaso
Dear all, 

I was wondering if it there could be a community interest providing a Gpars implementation for Reactive streams programming model, based on the Dataflow abstraction. 

The idea essentially is to write a stream data structure, by using dataflow queues, and complaint with this project, that I'm sure you know: 




I know that they are two different data abstractions but I tend to think the Reactive streams model as a special case of the Dataflow paradigm. 

In support of this, in Nextflow I've managed to implement operators semantically very similar to the ones provided by the RxJava framework (even though it does not pretend to be semantically equivalent to it) by simply using the Dataflow variables and queues. 



Since I was taking in consideration to make Nextflow compatible with the Reactive steams TCK, I would be interested to know if there's in the Gpars community any plan about it or in any case if there's a community interested on this. 


Best,
Paolo



Reply | Threaded
Open this post in threaded view
|

Re: Gpars dataflow and Reactive streams

Vaclav
Administrator
Hi Paolo,

not having received any comments on this, it seems there are no strong opinions on this in the community.
I feel like the GPars' Pipeline DSL as well as Dierk's Kanban-flow aim at a similar target - providing fluent API to wiring Dataflow channels with some form of back-pressure. An initiative that would lead to unifying and perhaps standardizing these APIs would definitely be worthwhile, in my opinion.

Cheers,

Vaclav





On Wed, Apr 30, 2014 at 2:17 PM, Paolo Di Tommaso <[hidden email]> wrote:
Dear all, 

I was wondering if it there could be a community interest providing a Gpars implementation for Reactive streams programming model, based on the Dataflow abstraction. 

The idea essentially is to write a stream data structure, by using dataflow queues, and complaint with this project, that I'm sure you know: 




I know that they are two different data abstractions but I tend to think the Reactive streams model as a special case of the Dataflow paradigm. 

In support of this, in Nextflow I've managed to implement operators semantically very similar to the ones provided by the RxJava framework (even though it does not pretend to be semantically equivalent to it) by simply using the Dataflow variables and queues. 



Since I was taking in consideration to make Nextflow compatible with the Reactive steams TCK, I would be interested to know if there's in the Gpars community any plan about it or in any case if there's a community interested on this. 


Best,
Paolo






--
E-mail: [hidden email]
Blog: http://www.jroller.com/vaclav
Linkedin page: http://www.linkedin.com/in/vaclavpech