Do you really need CouchDB in particular for this? Any Berkeley DB-style key-value pair database would excel at this.
Languages like Python have high-level wrappers that let you access such databases like a normal hash:
wordpairs[foo] = wordpairs[foo] + k # reads from disk then writes
In my experience you can get really fast simple persistence with basically no code this way. It's not as fast as can possibly, but then again the overhead of communicating via JSON and HTTP isn't, either.
Well, of course, with a language this sort of thing is possible. The beauty of the method described in the article is that it's built in the database system itself. No "language" is required (besides simple javascript for a view).
No one said couchdb was required. This article is simply showing how easy it is to accomplish incremental reduces.
Languages like Python have high-level wrappers that let you access such databases like a normal hash:
In my experience you can get really fast simple persistence with basically no code this way. It's not as fast as can possibly, but then again the overhead of communicating via JSON and HTTP isn't, either.Maybe I'm missing something.