Are you quite sure this actually speeds things up in practice? I tried this before and was disappointed in the results. The JVM is pretty fast at getting rid of newborns, and you're adding in the overhead of synchronization, twice. I'd like to see some empirical data on projects of different sizes.
http://gwt-code-reviews.appspot.com/34825 --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---
