A progress report can be found here. Summary: code is done, regression tests are ok, but I will spend some extra time to see if I can optimize somewhat without complicating the code too much (current implementation is 5 - 10% slower for the the metric that was the only one possible in the previous node). I think that a few percent is a reasonable price to pay for the extended functionality but if it is simple enough to tweak it, it would be wasteful not to do so 
This looks great!!! I would call that voronoi feature complete. Thanks a million for this! The slowdown is perfectly acceptable when compared with the dismal OSL performance (which I never use for that reason). With almost all of my cycles renderings I use the GPU, spare the occasional scene that seems to need absurd amounts of ram (usually 40+ GB).
Here is a greate overview collection of pattern and noises:
http://www.neilblevins.com/cg_education/procedural_noise/procedural_noise.html
Worley Voronoi F2-F1 is very ineressting…
varkenvarken, any procedural pattern you can squeeze in is great news, extended voronoi most of all. i hope your patch makes it to the master.
anyway, could you share any builds? would love to take it for a spin.