DESCRIPTION

Changed logging to slf4j

Change-Id: Ie293fa5c1da8474e419ae35242e87cf2e711147e


COMMENTS

author: Noah Dean
date: 2011-09-16 21:04:03.722000000

Patch Set 1: I would prefer that you didn't submit this

(6 inline comments)

Unless escape analysis got really good recently, I think a lot of this is going to be a lot more expensive than what it's replacing.  Much of it seems unnecessary.  It looks like the same type of features are available in slf4j.

(I don't have much of an opinion on slf4j in general)

I don't much like that every file has to have two lines of coded added to it at class load time to initialize a logger.  I imagine that may be a performance benefit in some places, but putting almost the same two lines of code in every file really feels like something isn't being done correctly.

Line:306, src/main/java/net/spy/memcached/MemcachedConnection.java -> This one is likely to be a lot more expensive.

Line:433, src/main/java/net/spy/memcached/MemcachedConnection.java -> I expect this to be very expensive.

Line:87, src/main/java/net/spy/memcached/CouchbaseNode.java -> I realize this isn't a new change, but we shouldn't be calling e.printStackTrace anywhere.

Line:229, src/main/java/net/spy/memcached/CouchbaseNode.java -> I really think most cases of ex.getMessage() are wrong.  Can't you pass the whole exception to the logger?

Line:49, src/main/java/net/spy/memcached/protocol/binary/OperationImpl.java -> I've never been a huge fan of this.  It requires putting *almost* the same code in every file.

Line:131, src/main/java/net/spy/memcached/protocol/binary/OperationImpl.java -> This is a pretty negative change.  You're constructing a string where, in the general case, you just throw it away.  I would expect this to have a noticeable performance impact throughout the code.

-------------------------------------
author: Kenny Maynard
date: 2011-09-26 23:07:54.210000000

Patch Set 1: Abandoned

Not going to change the loggin scheme

-------------------------------------
author: Andreas Short
date: 2012-12-11 01:27:13.734000000

Patch Set 1:

Just to add more context to this abandon, the other issue was that it changed behavior which we do not want to do in a 2.8.x release.  We want to change to slf4j, but can't have a .micro release suddenly start logging somewhere else by default.

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