DESCRIPTION

Removed all Membase and Couchbase support

We are splitting up this project into three seperate projects.
Each project will support a specific database.

Change-Id: I383dfcde9b5c1fbd0225e5a5e8ee001e8e3ada60


COMMENTS

author: Kenny Maynard
date: 2011-11-22 20:56:01.058000000

Uploaded patch set 2.

-------------------------------------
author: Andreas Short
date: 2011-11-29 04:25:30.611000000

Patch Set 2: Do not submit

I'm not inclined to support this series of commits.  We have lots of ripping out of code followed by commits that remove the build/testing description for that code.  This means this whole set would be a bunch of stuff that wouldn't build.

These need to be reordered and squashed into something that makes more sense.

Have sent an email about this directly last week as well (11/25).

-------------------------------------
author: Andreas Short
date: 2011-12-05 21:13:34.467000000

Patch Set 2: Looks good to me, but someone else must approve

(5 inline comments)

I'm going +1 on this as I think I'll need to see the other side.  I can generally imagine how it'll work, but there's some confusion for me with the ketama node locator and some of the other reconfigurables.

Line:64, src/main/java/net/spy/memcached/MemcachedConnection.java -> Hm, now I'm curious how this will come together.  I guess by just extending it?

Line:146, src/test/java/net/spy/memcached/QueueOverflowTest.java -> Also doesn't belong in this commit, but it's okay.

Line:9, /COMMIT_MSG -> I think it'll actually be two separate projects.

Line:98, src/main/java/net/spy/memcached/AddrUtil.java -> doesn't really belong in this commit, but it's okay.

Line:154, src/main/java/net/spy/memcached/KetamaNodeLocator.java -> How will this work in situations where we do want to reconfigure the KetamaNodeLocator.  I think I'm going to have to see the other side of the commits.

-------------------------------------
author: Kenny Maynard
date: 2011-12-05 21:26:06.263000000

Patch Set 2: (3 inline comments)



Line:64, src/main/java/net/spy/memcached/MemcachedConnection.java -> Yeah. I think this is really what we should have done in the first place. It also makes shutting down the client much easier.

Line:98, src/main/java/net/spy/memcached/AddrUtil.java -> Yeah, but I want to leave if for now. The reason is that the Couchbase project would have to have it's own separate AddrUtil class (maybe CouchbaseAddrUtil). I also left a few operations in Spymemcached that memcached doesn't support (like getl). In the future I think we will have a better idea on what the right thing to do is.

Line:154, src/main/java/net/spy/memcached/KetamaNodeLocator.java -> From MembaseConnection:

      if (locator instanceof VBucketNodeLocator) {
        ((VBucketNodeLocator)locator).updateLocator(mergedNodes,
            bucket.getConfig());
      } else {
        locator.updateLocator(mergedNodes);
      }

This to me is better since all of those other NodeLocators don't really have any idea of what a Config is. I remember when the change that did this went through and I personally didn't like it, but couldn't propose a better fix at the time.

-------------------------------------
