Turns out the reduce is really the problem. Well, not really the reduce but the map function. The library I am writing was designed mainly to store strings other simple types. But since CouchDB can store complex types, people wanted to store them and I just said 'yeah, you can do that'...

Obviously those complex types are mapped and computed in the reduce which terribly slows down the indexer.

Thank you Adam for pointing me into the right direction!

Daniel

Reply via email to