On Jan 15, 2008 5:01 PM, Jean-Sebastien Delfino <[EMAIL PROTECTED]> wrote:
> Simon Nash wrote: > [snip] > >>>> a) runtimes of various kinds (SCA standalone, embedded within Tomcat, > >>> etc) > [snip] > >>> b) applications, containing only the code and other artifacts > >>> required for the application itself > [snip] > >>> Drop contribution jars into a directory which are then picked up > >>> automatically and all contained composites are run > [snip] > >>> Webapp > >>> Construct a webapp > >>> Add dependencies (using maven?) for tuscany jars > >>> Configure web.xml to include the Tuscany filters > [snip] > > I've been thinking about how to provide useful input here but I'm still > a little puzzled as it mixes different topics in one thread. I've left a > few snippets above to try to illustrate that, and there's more (about > distro packaging for example) in other responses to the thread. > > Here are the topics I think I've seen: > > - Does the app developer need to know what Tuscany and dependency JARs > are required by his SCA contributions? > > - Under which circumstances does the app packager want to package the > Tuscany and dependency JARs with the application artifacts. > > - Which containers are we supporting and what kind of integration with > them? Tomcat is just one example, Geronimo, JMS providers, embedded > Jetty are others. > > - Does the app developer need to know ahead of time which container > integration his app is going to be deployed to. For example can I use > the EJB, WS and JMS bindings together in one app, and which container > integration is going to support that? > > - What distro Zips are we building and what do they contain? just the > runtime? samples or not? dependencies or not? are we building > specialized distros for different use cases? > > Would it make sense to spawn separate threads for these different topics? > -- > Jean-Sebastien > > All these topics seem related to me so so far i think it helps having the one thread about them. Not every email needs to be about them all, just snip out text leaving the bit you want to comment on, threaded news readers make it easy enough to keep track of the discussion. I'd rather have you participate though so do feel welcome to start other related threads. I think we need to keep in mind the bigger overall picture when thinking about these details, if we've been getting too much into the implementation detail would help to step back a bit - the following are the things i think we're trying to do here: 1) applications to contain only the code and other artifacts required for the application itself not Tuscany internals - simple sca contribution jars 2) some sort of runtime(s) which can run those application contribution jars, so that could be things like standalone (command prompt), from testcases, in a webapp, or some customization of something like Tomcat/Geronimo/WebSphere/JBoss etc I think we also need: 3) Those runtimes need to be distributed but the Tuscany download page should have a small number of downloads so its easy for users to work out which to choose 4) To make it easy for newbies to run things so at least one distribution should have prebuilt samples so they can be run out-of-the box without having to be compiled/built and for things to work without having to do much/any messing about with installation/configuration/customizing. 5) Ideally the easy to use distribution is not so large. Do all those sound reasonable? Does anyone want to add/remove/modify anything? ...ant
