my initial RDF/SPARQL-backed web apps are an experience in latency:

   ⎗
   web client generates some structure encapsulating user parameters
   client serializes JSON or XML
   client sends over network ⇢
⇢ app server parses JSON
   app generates some SPARQL
   app sends this to SPARQL endpoint over network ⇢
⇢ endpoint parses SPARQL
   endpoint generates SQL
   endpoint sends SQL over network to daemon ⇢
⇢ daemon parses SQL, optimizes/prepares query
   daemon runs query, serializes results
⇠ daemon returns results over network
   endpoint recieves query results, possible multiple SQL roundtrips to finish SPARQL query ↺↺↺
   endpoint serializes SPARQL results
⇠ endpoint returns results over network
   app parses results
   app generates HTML or JSON
⇠ app returns HTML or JSON over network
   client parses JSON
   client updates DOM ⎘

i need ruby or comparable for the vast library of building blocks and tools, i dont want to require javascript, since its disabled in my browsers, plus i use console mode tools and want accessibility and JS/DOM runtimes are so exceedingly large and complex at this point that only a few companies like microsoft, google, and mozilla are able to maintain them. i want more semantics than key/value stores, and i want it to run on my phone, which rules out the massive stores designed for web-scale mainframe approach (since ads = potential profit but you have to get users coming to your mainframe and therefore store all their stuff in some massive DB). i dont even want a database engine running at all, if possible - Gentoo and Git are using the filesystem just fine to get the job done

so, nobody's making what i need, and it doesn't exist…