Are you thinking you'd prefer the deployer to be part of the tck, or a module in its own right?
I don't have a strong view either way, as long as the TCK setup is doing the right thing. Maybe our Arquillian adapters should be using this method as well? Jon On Nov 7, 2011 7:15 PM, "Romain Manni-Bucau" <rmannibu...@gmail.com> wrote: > ok but the webappdeployer is not linked to the deployment at all, just to > the way we deploy tcks so should we keep this ejb on trunk? > > - Romain > > > 2011/11/7 David Blevins <david.blev...@gmail.com> > > > > > On Nov 6, 2011, at 4:22 PM, Jonathan Gallimore wrote: > > > > > My understanding - and this might be wrong - is that the existing > > > DeployerEjb behaves differently to just dropping the .war file in the > > > webapps directory. When we deploy using DeployerEjb, OpenEJB processes > > the > > > .war file first, and then hands it off to Tomcat. Dropping the war in > the > > > webapps folder on the other hand, means Tomcat processes the .war file > > > first and then OpenEJB gets its turn. > > > > > > I think the goal here is for the TCK to be as close to dropping the > > > archives in the webapps folders as a user would do as possible. > > > > Right. Close as in identical :) Unless we're going to tell people > "don't > > drop apps in webapps/", we should test it. > > > > Next step is to get the VmDeploymentManager class updated so the impl can > > be configurable. Then update the "runtests" script so the implementation > > cab be set for a test run. Then of course to try a test or two to verify > > that all works. > > > > Then we can kick off an entire run with that approach and see where we > > land. > > > > > > -David > > > > > > >
