|
From: Doug C. <cu...@ap...> - 2008-03-05 22:41:53
|
Doug Cutting wrote: > We might get started by using threads and method calls instead of RPC. > This should help us get our design straight before we invest in a "real" > implementation. So we might: > > 1. Write a simple client API: addDoc(), removeDoc(), updateDoc(), query(). I finally committed the abstract API. I'm out of town Thursday and Friday, but with luck may get some time to start a naive, in-memory, synchronouse implementation that, e.g., uses Java Collections as its index, that we can test against. > 2. Write a multi-threaded test program that uses this, simulating a > large, active index. It randomly adds, removes and updates randomly > created documents, and periodically queries and checks that results are > correct. I hope to start this next week. (That line sounds familiar...) > 3. Implement the "distributed" system using threads per node. I think last week's discussions gave us enough of a design to start implementing this. Should we attempt to summarize that design anywhere before trying to code it? Sorry this has been so slow to get started! The good news is that I've cleared lots of other stuff off my plate recently and hope to spend more time on Bailey. Doug |