On Tue, Apr 19, 2011 at 5:03 PM, Simon Laws <simonsl...@googlemail.com> wrote: > I committed a new version of two-nodes-two-vms-hazelcast. I called the > new one two-jvm-hazelcast to match up with the existing one-jvm-* > names. I'll remove the old one shortly. The new one's not in the build > as it doesn't shut down properly. I was though pleasantly surprised > that the distributed domain pretty much worked off the bad. Nice work > Ant for sorting that out again. > > In creating the new version I decided that I wanted to look at a > simplistic model for representing a domain on disc. For this test I > simply did... > > http://svn.apache.org/repos/asf/tuscany/sca-java-2.x/trunk/testing/itest/nodes/two-jvm-hazelcast/src/test/resources/domain-domain1/ > > I created the simplest piece of code possible to read this > > http://svn.apache.org/repos/asf/tuscany/sca-java-2.x/trunk/testing/itest/nodes/two-jvm-hazelcast/src/main/java/org/apache/tuscany/sca/impl/Tuscany.java > > I ended up finding that I was pretty comfortable about having a domain > on disc and then starting nodes with minimal configuration, I'd like > to look at working this up into a real piece of code in the shell or > the launcher. Anyone have any thoughts about this. >
You can use the Shell load command with one of those node.xml files. The name "load" to match the TuscanyRuntime loadFromXML method. I've just updated the startDeployables attribute to default to true which answers the comment in the Tuscany class about how to start it, and that also means it no longer needs the explicit <deploymentComposite> element. I'm not sure how well it will work using Ant tasks to run the two processes, from what i remember that approach always ended up causing intermitent hangs in the build, which i think is one of the reasons why all these type tests have ended up taken out of the build. ...ant
