Atlassian Blogs

Comments (7)

• 

  Hi,
  Tried the tool as well and find it hardly useful, really. Our picture is similar to yours (meaning it is as useless as yours) which, practically, says me nothing ..

  By Evgeny Goldin on August 30, 2007 / Reply

• 

  I was impressed with depgraph, which is a maven2 plugin from the el4j project. It uses graphviz to generate a slightly more readable (if considerably larger) dependency graph.
  It’s added in the http://svn.atlassian.com/svn/public/contrib/repository/maven2 repo, so all you need the following in your POM and graphviz and specify the mvn target as depgraph:depgraph.
  As I can’t paste XML here, here’s a link to an
  example pom.
  You just need to copy the ch.elca.el4j.maven.plugins group into your poms build section.
  The Images come out wide, but legible, usability may depend on the number of widescreen lcd’s you have Here’s the Example image (save to desktop, rendering in a browser is useless.)

  By Andy Brook on September 2, 2007 / Reply

• 

  Don’t be proud of this. Atlassian products are bloated memory hogs. Consider reducing the number of dependencies instead of boasting about them.
  e.g. Crowd is 154MB download for what is a pretty simple application. That’s disgusting and doesn’t reflect well on Atlassian.

  By Adrian on September 8, 2007 / Reply

• 

  Thanks for the feedback, Adrian. We are continually trying to improve both the performance and memory use of our applications.
  What I was trying to convey with a bit of wry humour is how many dependencies Confluence has, and how this is a bit out of control. We definitely need to get a better handle on what libraries we’re including, and reduce the number wherever possible. In fact, the exercise I was undertaking was to remove as many duplicated classes as possible from the classpath. (This doesn’t necessarily reduce memory usage, but it does have an impact on the download size as you’ve noted.)
  As a related matter, we’ve had a position open for several months looking for a J2EE performance engineer. If you have any experience with tuning the memory usage or performance of Java web applications, or know someone who does, please get in touch!

  By Matt Ryall on September 9, 2007 / Reply

• 

  Hi Evgeny,
  The purpose of the dependency analyzer is not just showing the dependency graph of your project, but to help you dealing with maven2 related issues such as solving duplicated dependencies, realizing the source of unwanted dependencies, understanding the origin of your dependencies version source and more.
  This can be done by filtering your dependency graph according to dependency scopes, dependency groups ids (good for concentrating on a specific issue), grouping similar dependencies and more.
  Also, it is possible to drill down into a specific dependency and view its dependencies in another tab.

  By Dror Bereznitsky on September 11, 2007 / Reply

• 

  While I started off thinking that depgraph might help, and agree that it would be useful seeing the complexity, I think in addition to using depgraph developers should just check the WEB-INF/lib directory periodically to make sure you don’t have duplicate jars or other jars you might be able to exclude in the pom. I just raised this issue here for Confluence if you want to track it and respond there:
  http://jira.atlassian.com/browse/CONF-9760

  By Gary Weaver on October 18, 2007 / Reply

• 

  Gary, we’ve implemented a couple of automated tests to check this going forward. I’ve responded with more details on the issue.

  By Matt Ryall on October 22, 2007 / Reply