The other option we have it setting a parentBeanFactory in the application scope, in the application.cfc and setting it as the parent bean factory, and having each machii application point to an empty coldspring XML config file, which seems to be a bit of a workaround.
I actually recommend that option if I am understanding your situation correctly. The problem you will probably run into is that each Mach II app in your application scope needs access to the services defined in your coldspring config file. Your CS config file right now is loaded the first time each application is hit for the first time. Since all your apps need to share the same bean factory the best option would be to initialize that bean factory in the applicationStart method of application.cfc and place it in the application scope. Then have each of your Mach II apps pick that bean factory up as the parent bean factory. As your apps grow they can then override or use parts of the application wide bean factory you setup as a parent. --Kurt On 12/17/06, Mark Mandel <[EMAIL PROTECTED]> wrote:
Another question to you boys about the 'placeFactoryInApplicationScope' attribute. We have multiple machii applications within our system, all sitting under the same Application.cfc, and they all use the same ColdSpring file. I had assumed that if we set placeFactoryInApplicationScope=true on all our machii applications, they would check to see if ColdSpring was already loaded in the application, and if it was, simply pick up the beanfactory that was already stored in the application scope, and use that. Instead it seems to load a new CS beanfactory, and overwrites the one in the application scope. Is this the way it is meant to be? The other option we have it setting a parentBeanFactory in the application scope, in the application.cfc and setting it as the parent bean factory, and having each machii application point to an empty coldspring XML config file, which seems to be a bit of a workaround. I can see that there are hooks in place in the BeanFactoryUtils.cfc to check for the existence of named factories, and the functionality simply becomes a check to see if placeFactoryInApplicationScope is true, and if the bfUtils.namedFactoryExists(), then get the namedFactory from the bfUtils and set it to the loca version. So.. is this something rediculous, or just something we just haven't talked about before? Mark -- E: [EMAIL PROTECTED] W: www.compoundtheory.com
