test.rb

#!/usr/bin/env ruby 

require 'rdf'
require 'rdf/raptor'

repo = RDF::Repository.new

count = 0
while true
  repo.load("test.ttl")
  count += 1
  puts count
end

test.ttl

@prefix rdfs: <http://www.w3.org/2000/01/01/rdf-schema#> .

<http://example.org/1> rdfs:label "Test" .

From the trace in http://github.com/bendiken/rdf/issues/#issue/33/comment/294533 it looks like rdf/raptor is falling back to the cli implementation, which happens if it cannot locate the libraptor library. Try setting the RDF_RAPTOR_LIBPATH environment variable to the full path of your libraptor library and see if that solves the problem. That doesn't fix what appears to be a bug in the cli implementation, but the ffi implementation is faster.

It works great and makes faster after setting RDF_RAPTOR_LIBPATH=/opt/local/lib/libraptor.dylib. Thanks! I should add the /opt/local/lib directory into the library path anyway...

Please do file a bug against rdf-raptor for the cli implementation though, referencing this and the issue filed against rdf. Thanks.

will do

noted at http://github.com/bendiken/rdf-raptor/issues#issue/6

I suppose we could also consider improving rdf-raptor's CLI/FFI mode autodetection to look for the FFI library in some well-known locations such as this one. For instance, most Mac OS X users would benefit from it trying the MacPorts location of /opt/local/lib, yet that path will rarely be in the system's library path since that is known to cause conflicts and general weirdness (hence the numerous DYLD_FALLBACK_LIBRARY_PATH hacks out there, and so on). Something I'll try and keep in mind for the next point release (please feel free to open a ticket if you'd like to see this feature).