Spring Transactions, Hibernate Mappings, and JRuby Growth 
A recent poll of the Spring community produced some interesting results. The question was "How do you limit transaction boundaries?" and I was surprised that only 3.5% responded that they are using the BeanNameAutoProxyCreator. See for yourself at
http://forum.springframework.org/showthread.php?t=37178 .
Today I had an interesting observation about the placement of Hibernate mapping files in a multi-layer system. I noticed that the mapping files have historically been placed in the same project (physical JAR) as the domain objects they support. However the rest of the Hibernate implementation goodness is taken care of in a separate data (i.e. persistence) layer, subsequently details of our ORM solution are now leaking through to every layer that has access to the domain objects. I feel the Hibernate mappings should stay in the data layer and just perhaps mimic the domain object's package namespace. Interestingly enough in the book "Hibernate Quickly" (by Patrick Peak and Nick Heudecker) it is suggested to keep your mapping files in the "same JAR file as the classes they describe." In the end this isn't going to make or break the software but I believe, at least in a multi-layered system, these files should be bundled with persistence code.
JRuby 0.9.9 has been released and two new committers have been added. Read all about it at
http://on-ruby.blogspot.com/2007/04/jruby-099.html .
