Am 16.01.2012 16:14, schrieb Raul Miller: > By "ipc" I think he means what I think is Q's . > > In Q, the "natural representation" of any item is a serialized version > -- evaluating it will recover the original item. This is not the case > in J -- for example 99x gets displayed as 99 but: > > 99 -&(^~) 99x > _3.98353e182 > > Anyways, if I understand Q properly (or maybe it was K), it will ship > a sentence off to another interpreter using . and the result is the > result from that other interpreter. And, even if I do not have the > syntax exactly right, the underlying point is that Q/K it's fairly > simple to delegate processing to a small farm of machines. This is the tick data feature available in kdb+. Very cool stuff. > This can > be useful, for example, when very large (multiple terabyte) data > structure are spread out across multiple machines. > > I believe that the usefulness of this ties in with Q's support for > tree data structures as well as triggers and dependencies. Can you here be more specific? There is an article about tree data structres in Q. But Q does not support it directly. Moreover triggers and dependencies are also not directly supported in Q. > > J does not currently have anything like that. And, for that matter, > 5!:5 does not always serialize in a form that ". can digest. >
---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
