I wanted something easy to pick up -- Javascript seemed natural. I also wanted to be as agnostic as possible, because graph query languages are interesting and worth experimenting with. That's why it has submodules per-language and it's (relatively) straight-forward to write another.

Cypher's not bad either, but it's not "just Javascript". But I'm totally taking commits if someone wants to port it :)

I'm a developer on Orly and I focus on the language portion of the project. Orly takes a different approach in this regard by providing a general-purpose functional language as its "query" language, rather than providing a language specifically designed to traverse/query the graph. This approach allows us to perform not only queries but also arbitrary computations on the server-side and take advantage of the powerful resources available on the server such as large memory, CPUs and perhaps most interesting are GPUs.

Getting performance out of a graph often depends on being able to express your query effectively. The last thing you want to do is plunk down a cursor on a node and have your client start wandering around the graph, following edges. All that back-and-forth chattiness is a non-starter network-wise and it gives the database engine essentially no chance to optimize the query. I've been favoring scripts that let me use free variables--like Prolog logic variables--to describe shapes in a graph and then let the database server find bindings that match. Like 'print a.name, b.name for a, b, x, y where a is_friend_of b and a is_friend_of x and b is friend_of y and not x is_friend_of y and x.city == y.city'