On 9/17/07, Jean-Sebastien Delfino <[EMAIL PROTECTED]> wrote: > > Jean-Sebastien Delfino wrote: > > Simon Nash wrote: > >> > >> Simon Laws wrote: > >> > >>> On 9/16/07, ant elder <[EMAIL PROTECTED]> wrote: > >>> > >>>> On 9/16/07, Simon Laws <[EMAIL PROTECTED]> wrote: > >>>> > >>>> <snip> > >>>> > >>>> A question though is should we flip the samples over to use NodeImpl > >>>> instead > >>>> > >>>>> of SCAdomain? > >>>> > >>>> > >>>> To save digging around in the code / ML, what is the new way exactly? > >>>> Currently things do: > >>>> > >>>> org.apache.tuscany.sca.host.embedded.SCADomain.newInstance(" > >>>> helloworldws.composite"); > >>>> > >>>> what is the new replacement approach? > >>>> > >>>> ...ant > >>> > >>> > >>> > >>> > >>> org.apache.tuscany.sca.node.impl.NodeImpl node = new SCANodeImpl(); > >>> node.start(); > >>> node.getContributionManager().startContribution(The URL to your > >>> contribution); > >>> > >>> For example, for test purposes the URL could be: > >>> > >>> SCANodeUtil.findContributionFromComposite(???.class.getClassLoader()," > >>> helloworldws.composite")); > >>> > >>> So it's more lines but removes some of the mystery about where the > >>> contributions are found. > >>> > >>> Simon > >>> > >> I don't like this at all. Is there a problem with staying with > >> what we currently have until we can design a clean replacement API? > >> > >> Simon > >> > >> > > > > The non-distributed samples could stay with the current > > o.a.t.embedded.host.SCADomain. > > > > The distributed samples need to move to a cleaner API. > > > > I made a proposal for a smaller and cleaner API for use in the > > distributed samples there: > > > http://svn.apache.org/repos/asf/incubator/tuscany/branches/sca-java-1.0/modules/node-api/src/main/java/org/apache/tuscany/sca/node/smaller/ > > > > > > However, I'm not too happy about having two different APIs for > > non-distributed and distributed... > > > > Another option would be to port the distributed samples to the > > existing o.a.t.embedded.host.SCADomain, hook the existing SCADomain > > with the distributed Node support, and introduce the Node API only > > post 1.0. > > > > -- > > Jean-Sebastien > > > > More thoughts, trying to cleanup the calculator-distributed API usage > while minimizing changes and impact on everything else as time to cut > 1.0 RC2 is approaching :) > > We could do the following: > > - Keep o.a.t.sca.host.embedded.SCADomain unchanged. > > - Copy it into a new o.a.t.sca.domain.SCADomain, in a small domain-api > module which will only contains that class and only depend on the > sca-api module. This becomes the API to use to leverage the distributed > domain support. > > - Add one parameter to one of the > o.a.t.sca.domain.SCADomain.newInstance() methods to take an additional > nodeURI parameter (as it's needed by nodes to tell the distributed > domain where they are). > > - Move the SCANode interface and its friends to module node and we can > probably remove the node-api for now. > > - Port the calculator-distributed sample to o.a.t.sca.domain.SCADomain. > > Thoughts? > > -- > Jean-Sebastien > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > sounds ok to me
Assuming this works we then flip o.a.t.s.host.embedded.SCADomain to the o.a.t.s.domain-api where ever it appears Simon