Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Received-SPF: pass (nike.apache.org: domain of randall.leeds@gmail.com
 designates 209.85.210.180 as permitted sender)
MIME-Version: 1.0
Date: Thu, 8 Dec 2011 16:04:38 -0800
Message-ID: 
 <CAAL6JQgPYz3h6oov3oy6cMVox2RQkvF0KaPobyuLwQc1+A3VLw@mail.gmail.com>
Subject: BTree Refactoring and Performance
From: Randall Leeds <randall.leeds@gmail.com>
To: dev@couchdb.apache.org
Content-Type: text/plain; charset=UTF-8

I just went back through COUCHDB-1084 [1] and COUCHDB-1124 [2] and it
seems like we hit a roadblock at benchmarking.

Do we need to do anything other than put these through their paces
against a ram-disk couch?

As a broader question about benchmarking and performance tracking:
Should we be doing anything with the ./test/bench directory? Are there
tools any of us are using out-of-tree that should be in-tree?
It'd be really amazing if someone stepped up to set up a build server
that runs a benchmark suite after every commit.

I know, I know. I'm firmly in the camp of people who don't care too
much about raw performance, and you may be, too. That doesn't mean
we're okay with performance regressions and, at least in these two
cases, fear of said regressions seems to be holding some nice patches
up.

-Randall