I found many blame Maven for many things that are not right, they just have not found the way to solve them.

Here it goes some of them:

• The learning curve is as steep as it can get, the documentation is rudimentary and limited and public domain is not much help either
  
  Almost all questions in the user list are answered, Maven has a growing community with lots of users. And once you are used to a Maven project you are used to all Maven projects all over the world.
• Limited Utilities to Integrate with existing IDEs like WSAD (eclipse 2.1 platform based)
  – mevenide is improving with time, anyway I don’t really need to use it, I prefer doing by hand.
• Tedious process of creating XDoc files
  – Use html2xdoc plugin
• Dependency management is a pain
  – Dependency management is one of the best things of Maven. You can’t blame maven because your alternatives are: doing it by hand writing lots of xml lines!
• Limited Integration facilities with SCM tools like Sub-Version, Serena etc.
  
  – It’s already done for next versions

I have just implemented a tiles controller that sets css styles based on user-agent header automatically.

The layout of the webapps can be easily customised without touching any jsp code nor creating new jsp files, CSS stylesheets are profusely used, no annoying tables, no need for typing different urls according to the browser you’re using, you can check that most of sites in the net are using different urls for pdas and other browsers, e.g. in Google you need to go to www.google.com/palm to see the palm site.

Here you can find the controller documentation
http://oness.sourceforge.net/multiproject/oness-common-webapp-controller/

And here some snapshots and the view documentation and tiles configuration file with a sample of setting a custom layout to a Palm pda.
http://oness.sourceforge.net/multiproject/oness-common-webapp-view

Today I’ve made major changes to the model layer. Hibernate is soooo complicated… Maybe it’s a good thing for simple applications but in a layered one there are many issues you have to deal with.

My idea is now:

• All collections are lazy to avoid sql queries
• Everytime you need a collection you have to call the DAO to get it, this way you also ensure independency of the persistence mechanism used
• When the DAO is called to get a collection it’s filtered to return only values not deleted (transactionTime.end is null)
• findById should return a clone of the object returned by Hibernate because it’s cached, so next modifications in facades don’t affect persisted instances.

Hope this works when rerunning test cases tomorrow morning. Then I’ll commit the changes.

http://www.springframework.com is now live!

Revenues from commercial activities will fund ongoing enhancements to the Spring Framework, including development, documentation, support, and testing. A substantial portion will directly fund the time of Interface21’s Director of Spring Development, Juergen Hoeller, to continue his level of commitment to the code base and user community.

The press release:
Interface21 To Provide Expert Spring Framework Consulting, Training, and Support

The guys from Hibernate have just made available an alpha version of Hibernate3, I’m looking forward to test it when I have a bit of time.

Hibernate3 is the world’s most sophisticated ORM solution. Building on Hibernate 2.1, the de facto standard ORM implementation for Java, Hibernate3 adds a wealth of new features including: “filters” for handling temporal, regional and permissioned data, support for mixing hand-written and generated SQL, new options for mapping exotic inheritance strategies and legacy data, runtime performance monitoring via JMX, and a new event-driven architecture. See the road map for more details and have a look at the Changelog.

The “filters” for handling temporal, regional and permissioned data looks great for the ONess project.

You can download Hibernate3 here.

Today we’ve improved the Maven central repository upload process, the time from a request to its upload to ibiblio will be much shorter from now. Hope this will make happy maven users.

After trying HSQLDB database it seems that doesn’t support setting transaction isolation levels, so it fails in testing.

I have adapted a bit the ONess application to work with postgresql database: changed the hibernate id generator native to increment, as dbunit test insert data but don’t update sequences, and renamed user table and columns because they are reserved words.

Maybe this helps people to better understand why to migrate to maven 😉