Proxy Agents!

[Linas Vepstas]

The AtomSpace now has working Proxy Agents. These can be used to wire up Storage to perform more complex data management tasks.

The most obvious example satisfies an old and previously unfulfilled request: how can multiple users share a single giant read-only dataset? This request came from the MOZI team, who had large genomics and proteomics datasets. Here you go. https://wiki.opencog.org/w/SequentialReadProxy

Another fun example involves caching. This can be used to provide a small AtomSpace window into some giant AtomSpace that is too big to live in RAM.  Another interesting use is to provide Episodic memory, so that the currently "hot" atoms live in the AtomSpace, but when attention has wandered away elsewhere, these get written back to disk.

In short, the proxy infrastructure provides shared, common wiring, and half-a-dozen example proxies. These are plenty enough to start wiring together some fairly interesting Atom-processing pipelines, blasting data around on the network, or building sharded large storage pools.  Here's the overview:

https://wiki.opencog.org/w/ProxyNode

This assumes some minor familiarity with StorageNodes in general:

https://wiki.opencog.org/w/StorageNode

A demo that works is here:

https://github.com/opencog/atomspace/blob/master/examples/atomspace/persist-proxy.scm

Happy Hacking 😀😀😀

-- Linas

--

Patrick: Are they laughing at us?

Sponge Bob: No, Patrick, they are laughing next to us.