Spaces - mehen

A space is mehen’s language-aware container for a function, method, class, trait, interface, module, or file. Spaces nest: a method lives inside a class which lives inside a file. Most metrics are reported per space, and aggregates roll up to the parent.

Why a single abstraction

Different languages call their containers different things — a Python class is not literally a Java class is not a Rust impl. But for metric purposes they all play the same structural role:

They contain functions/methods.
They define a scope for visibility (public / private).
They are addressable in tools and reports.

mehen’s Space abstraction unifies them so the engine, metric finalizers, and reporters can speak one vocabulary across languages.

Space kinds

Kind	Examples
`File`	Always the root space.
`Function`	Top-level functions, methods, SQL routines (a `CREATE FUNCTION`/`PROCEDURE`/`TRIGGER`, package-body routines, nested subprograms).
`Closure`	Lambdas / closures / arrow functions.
`Class`	`class`, `record`, `data class`.
`Interface`	`interface`, `trait`, `protocol`.
`Trait`	Rust traits, Kotlin traits where distinct.
`Impl`	Rust `impl` blocks.
`Module`	Files / Rust `mod` / Python `class`-as-namespace where applicable.
`Custom`	Analyzer-defined scopes for declarative languages — e.g. `sql.statement` (one top-level SQL statement; routine definitions nest as `Function` spaces inside it).

Aggregation

Every metric is computed at the leaf space (where it makes sense — a function for cyclomatic, a class for NPM, etc.) and then aggregated up to the file:

Sum for additive metrics (cyclomatic.sum, nom.functions).
Average for ratios (cyclomatic.average, nom.average_functions).
Min / max for distribution endpoints (cyclomatic.min, cyclomatic.max).

Aggregation rules live in mehen-metrics. Per-language analyzers decide which kind each tree node becomes.

See also

Concepts → Architecture — where spaces sit in the pipeline.
Code metrics overview — which metric is reported at which space kind.