I personally prefer the functional approach much better. And if you compare the word count examples on the respective readmes [1, 2], you will see riko is much more succinct. But I suppose the verbosity of the other libraries come with benefits like scaling across a cluster of servers.
We use somewhat different concepts. I tend to think of streams as infinite, so it didn't occur to me to include something like a reverse pipe operator.
I'm a bit surprised, why is filter an operator rather than a processor? I would think filters usually apply per-item, not to a whole stream?
I havn't worked on it very much but I'm heading towards push-based, using 0MQ for distribution/parallel processing, and using asyncio, mostly because it plays nicely with 0MQ.
We are in agreement. reverse has a notice that it isn't lazy [1]. I prefer to include pipes that aren't lazy since it can be helpful in some cases (plus the goal is to include all pipes originally in Yahoo! Pipes). The vast majority of pipes work just fine on infinite streams [2].
> I'm a bit surprised, why is filter an operator rather than a processor? I would think filters usually apply per-item, not to a whole stream?
Just an implementation detail [3]. I agree it would be better if it were a processor since it could be parallelized. PRs welcome :).
> I haven't worked on it very much but I'm heading towards push-based, using 0MQ for distribution/parallel processing, and using asyncio, mostly because it plays nicely with 0MQ.
See my previous comments related to this [4, 5].
[1] https://github.com/nerevu/riko/blob/master/riko/modules/reve...
[2] Assuming you're not using the async or parallel mode
[3] https://github.com/nerevu/riko/blob/master/riko/modules/filt...