incubator-cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <jbel...@gmail.com>
Subject Re: Random Checkins
Date Fri, 03 Apr 2009 02:21:22 GMT
On Thu, Apr 2, 2009 at 6:37 PM, Avinash Lakshman
<avinash.lakshman@gmail.com> wrote:
> Another issue with ConcurrentHashMap is that the dude is a memory hog. We
> got rid of it over a year ago because of the very same reason.

Re CHM: what do you suggest instead?  I assume you are talking about
for EfficientBidiMap.  Do you want the Getter to return a copy of the
ColumnFamily and stick with non-concurrent structures?  Or use
NonBlockingHashMap?

> Also for
> everything that is done there is a reason.

I don't see any harm in, for instance, me making a commit, you saying
"ConcurrentHashMap is a memory hog", and me saying, "okay, what do you
suggest instead?"  Nothing wrong with post-commit review in trunk.

> I think we are not asking for
> much to apart from run it by someone before what are basically changes that
> suit your style.

We've tried waiting for you or Prashant to review things.  That didn't
work over on the code.google project and it hasn't worked here.  My
remove patches sat unreviewed for weeks, literally.  More recently, a
request for a list of known issues and my attempt to start  a
conversation about a roadmap have gone without comment so far.

I'm not trying to dump on you; we know you're busy.  But we have
deadlines too and we can't hold every patch waiting days or weeks for
a review.

Tell you what.  I will try to get changes reviewed by someone in the
community before committing.  We will see how that works.  That should
at least give you some confidence that it's not just me off being a
cowboy.

-Jonathan

Mime
View raw message