On Tue, Nov 18, 2008 at 4:32 AM, Dave <[EMAIL PROTECTED]> wrote: >> I only got Roller + GlassFish working together, by adding the following JPA >> configuration to my roller-custom.properties: >> openjpa.QueryCompilationCache=false > > How did you figure that one out?
Well I knew it was possible that JPA configuration defined in roller-custom.properties will override what's in JPAEMF.properties - and the rest of it was just a lucky guess (that happened to worked first time around). I Googled for the existence of "QueryCompilationCache" and setting it to false did the trick. >> I've fixed it, and it works, but the problem is, I am not 100% sure why it >> works! Does this look like a bug in Roller? GlassFish? Anyone who could shed >> some >> light on this? I also need to understand the implications of turning off >> OpenJPA's QueryCompilationCache. > > I'm not sure what problem is, but it looks like some sort of > classloader craziness that could burn some hours to investigate. It's > possible that the problem does not exist in OpenJPA 1.x -- > unfortunately, Roller is still using OpenJPA 0.97. Yep, its certainly one of those classloader-type issues that could take up far too much time (that I don't really have at the moment). I think I read (perhaps it was in JIRA, or perhaps on here) that the work to upgrade to OpenJPA 1.x is non-trivial? Its more than dropping in the JAR? (I am talking about creating a custom build from source) > And I'm not sure what the performance implications of turning off the > query compilation cache. It's probably a bad thing, but I don't know > how much of a performance hit you'd take. Thanks. I haven't begun work to evaluate performance (there are a dozen or so other options that I also need to tweak, such as caching settings), but I'll post if it turns out to be a huge performance problem. For now, should I ticket this in JIRA?
