On Fri, Jun 6, 2008 at 1:56 PM, Miles, Chris <[EMAIL PROTECTED]> wrote:
> Hi, > > > > I have just downloaded Tuscany and have read through some of the > beginner's guides and some various pdfs and blogs regarding Tuscany. I > may have some wrong preconceptions about Tuscany, and maybe someone can > set me straight on it. > > > > I am unsure how I might go about setting up this possible environment. > The way I see my application working is multiple jar files each > containing a component which can be completely redeployed on-the-fly > much like the way I currently deploy EJBs at the moment. So say I have > just fixed some bugs on a ProductMgr component and I want to deploy this > out to the environment (either webapp or standalone). I want to be able > to deploy the webapp completely separately and then just be able to find > and load the component as needed, but I cant seem to see any > documentation explaining such a process. All the samples I have looked > at seem to be component based but are all deployed as a single unit. > > > > Any help would be greatly appreciated. > > > > Thanks > > > > Chris > > Hi Chris I'm not sure exactly what you mean by " want to be able to deploy the webapp completely separately and then just be able to find and load the component". The word "load" is what got me stuck. Do you mean find and "use" the component. Do you have in your mind that components are loaded on demand? Maybe you could say a little more about your mental model and I'll try and put that into context with how Tuscany is currently. I'm very interested in what people think is a good deployment model for SCA applications. You're right that many of our samples simply deploy and run all of the composites/components required all at the same time. There are a couple that don't do that. In particular sca/samples/calculator-distributed and sca/tutorial use our new domain manager application. This allows the SCA application to run on a set of nodes distributed across the network. You launch the domain manager before you launch any nodes that are actually going to run parts of the SCA application. You add contributions (which contain parts of your application) and you configure which nodes the composites from those contributions will run on. Once you are set you can run up the nodes and have them read their configuration from the domain manager and your application is then up and running. See the READMEs in the two directories and give them a go. This may not be quite what you had in mind as there is an expectation that in order to run an application all the parts of it are configured in the domain manager before you run the individual parts. It doesn't mean that all the parts have to be started at exactly the same time but it does mean that the doman manager has to be aware of them before you try to run them. Ant has done some work recently in the runtime-tomcat module when constructs a domain just be dropping wars into Tomcat. Early days for this though. I realize I've sneaked in a lot of terms here like Contribution, Composite, Domain, Node etc. so ask for me to explain further the parts that aren't clear. Regards Simon