One of the great features of 7 is the dependency analysis matrix. Probably, there's an internal API that is used to create it. Hopefully, when it's ready, OpenAPI developers will have access to that API.
In the meantime, we have the great example of com.sixrr.metrics.dependency. The techniques there to analyze dependencies appear to me to revolve around the PSI.
I am wondering if this is not also the way that the DSM is created, or is there some more efficient technique used by the DSM to analyze dependencies.
Is that right- all dependency analysis ultimately falls back on the PSI?