Tracer

Typeclass for algorithm tracing. Implementations produce human-readable output showing the steps of a graph algorithm at a given nesting depth; the depth is used to indent output so that the logical structure of the algorithm is visible.

Each trace call provides two lazily-evaluated messages:

• brief — a short summary (e.g. a count); always used by non-silent tracers
• detail — a fuller description (e.g. collection elements); used only by verbose

The canonical no-op given (Tracer.silent) is resolved automatically, so existing call sites require no changes. Students opt in by summoning a tracer:

 given Tracer[Int] = Tracer.verbose()
 Kosaraju.stronglyConnectedComponents(graph)

Both brief and detail are by-name so that string interpolation is never evaluated in the silent case — zero overhead in production use.

Three built-in modes:

• silent — no output, neither brief nor detail evaluated
• summary — emits brief only, and only at depth 0
• verbose — emits detail (falling back to brief if detail is empty), up to a configurable maxDepth