From: Anthony E. <me...@an...> - 2001-11-25 23:10:58
|
> > > > >> >> > >> >> You want to be able to reload the config at time intervals for one. > >> > > >> > Will do. I suppose this should be a paramter in the configuration? > >> > >> Yes. I'm making it an attribute of the top-level integradata element. > >> > > > > Nick, > > > > Actually, it already monitors for configuration changes. Right now it > > defaults to 5 seconds, but I will add the ability to configure it > > through a configuration field before the next release. This is > > different then reloading at specific intervals and only works with > > resources which maintain a modification time, but I feel that it is a > > useful feature none the less. Regardless, I should probably also > > provide a way to specify specific reload intervals as well as the > > ability to disable all of these features. > > > > -Anthony > > Yes, an optional attribute is what I didn't say: I'm influenced > by the same > forces you are. However, I'm not sure what you mean by "resources which > maintain a modification time", as I would assume an XML file would by > definition always subject to mod time changes, checksumming and all that. XML files may not always be the configuration source, although they are the most likely. However you conceivably pull your XML data from any data source which provides an InputStream. > I still do not have a brilliant idea to make an elegant way to plug my > rules into your framework except for an Ant build target that substitutes > import statements and all that jazz at compile time. This is > "release-only" > code. Not much value in the source itself. Yeech. > > The really interesting thing to me about any data validation framework is > high-quality rules-plugability, not merely a system to plug them into. So > in my mind this is of cardinal importance and impact for the long run for > both of us. So it is not possible for your classes to implement the org.formproc.validation.Rule interface? Their is no need to be in any specific package as the classes are dynamically loaded. Do all of your rules already implement a common interface? -Anthony |