I would be grateful for a sanity check of how we are thinking of using p2 while building the Eclipse Virgo project.
Virgo is split into a number of components each of which has a git repository and a Hudson build. The components depend on each other, with the dependencies forming a DAG.
When we successfully build a snapshot of a component, we want to make the bundles of the snapshot available for builds of other components to consume. We have been advised to use p2.
So we are thinking of making each snapshot build construct a p2 repository of built bundles in the build's workspace and, if the build is successful, to promote the contents of the workspace p2 repository to a common Virgo p2 repository in the Virgo downloads area.
The benefit of publishing to a common repository is that component builds need only specify the location of the common repository rather than having to construct some kind of search sequence of all the repositories of components containing dependencies.