GraphLint: Creating a Domain Specific Language for Graph Validation

Re:GraphLint: Creating a Domain Specific Language for Graph ... by alexholehouse on Wednesday February 08, 2012 @12:24PM (#3567426)

This is really cool - prolog and datalog are great for a wide variety of things, but I'd imagine the performance advantages writing a DSL rather than having to go through the whole *log interpreter is a major plus.

Out of interest, what's the performance like at scale (50 000+ nodes)? (In hind-site, I guess I don't actually have anything to compare it to, but a 50 000 node graph seems pretty big!)

[ Parent | on HN ]

Re:GraphLint: Creating a Domain Specific Language for Graph ... by jordanlewis on Wednesday February 08, 2012 @12:54PM (#3567584)

Kenny's initial implementation of the validation on a sample 50,000 node data graph and a specific ontology took four seconds.

The goal of Kenny's project was to get a working DSL and prototype of the validation algorithm. We don't really have much performance data to compare it to either at this point. As we gather that data by using the system, we'll make it faster as necessary. (Nathan Marz's "suffering-oriented programming" post is a perfect description of this development process http://nathanmarz.com/blog/suffering-oriented-programming.ht...)

[ Parent | on HN ]