Further info:
We are starting to scale in terms of team members and services, we have a dozens of completed PR’s per project, and continuously scanning long lived branch is taking up too many available agents.
We are looking to optimise our pipelines.
PR analysis’ ability to only analyze the changed files, i.e. incremental analysis relies on there being a cached knowledge base of the unchanged files on the server to fill in the gaps in the semantic picture.
So PR analysis itself doesn’t require a previous analysis of the target branch. But having one on file may speed the process up.
Actually, incremental analysis aside, it seems that an up-to-date analysis of the target branch is required to correctly detect which issues are “new”.