In short
Extra whole-project 'how tangled is everything' scores (standard numbers like CCD/ACD you can trend over time or compare to an ideal), on top of the per-file metrics it already computes.
Problem
Cumulative dependency metrics summarize how entangled a whole component graph is in a single comparable number (and how far it is from an ideal balanced tree) — exactly what you want to trend and gate over time. The existing metrics are per-element; these are whole-graph.
Proposal
Add cumulative component-dependency metrics — CCD, ACD, RACD, NCCD — and per-component visibility metrics on the existing metrics(...) entry, e.g. metrics("src/").cumulative().ccd().should_be_below(...), alongside the existing LCOM/coupling/abstractness/instability metrics.
Acceptance criteria
In short
Extra whole-project 'how tangled is everything' scores (standard numbers like CCD/ACD you can trend over time or compare to an ideal), on top of the per-file metrics it already computes.
Problem
Cumulative dependency metrics summarize how entangled a whole component graph is in a single comparable number (and how far it is from an ideal balanced tree) — exactly what you want to trend and gate over time. The existing metrics are per-element; these are whole-graph.
Proposal
Add cumulative component-dependency metrics — CCD, ACD, RACD, NCCD — and per-component visibility metrics on the existing
metrics(...)entry, e.g.metrics("src/").cumulative().ccd().should_be_below(...), alongside the existing LCOM/coupling/abstractness/instability metrics.Acceptance criteria