Tuple Spaces (Or, Good Ideas Don't Always Win) (2011)

It's comparable to FBP, and indeed, such a comparison has been made, specifically with Linda: [0]

In effect, the difference stems from whether you want to explicitly allocate ports to processing nodes, or if you want nodes that discern connections associatively.

9p/styx is also worth bringing into the discussion for an inverted understanding of the problem: where these languages suggest what the graph should look like when operating correctly, as in a batch process, what Plan9 did was to reason through how to connect up the nodes dynamically and debug multiprocessing "in detail", which is relevant when your connections fail.

[0] https://www.jpaulmorrison.com/cgi-bin/wiki.pl?CoordinationLa...

Previously

https://news.ycombinator.com/item?id=17635413

https://news.ycombinator.com/item?id=29213879

Python implementation: https://pypi.org/project/lindypy/

Really nice article! I keep coming back to the ideas of Tuple spaces too, and am sure it will become mainstream one day.