Missing Coverage Policy

When go-crap finds a function that has no coverage data at all, it needs a strategy. The --missing flag controls this behavior.

Note: "no coverage data" means the function does not appear in any coverage profile — typically because the package has no test file, or the function is in a file excluded from the test build. This is different from a function that appears in a coverage profile with 0% coverage (all statements present but none executed). Functions with known 0% coverage are always reported with their actual score, regardless of the --missing policy.

Options

Policy	Flag	Effect
pessimistic	--missing pessimistic (default)	Assume 0% coverage x maximum CRAP score
optimistic	--missing optimistic	Assume 100% coverage x minimum CRAP score
skip	--missing skip	Exclude the function from results

Pessimistic (default)

Assumes functions without coverage have 0% test coverage. This gives the highest possible CRAP score and surfaces them as top issues.

go-crap scan --missing pessimistic

A function with CC=8 and 0% coverage:

\[ CRAP = 8^2 x (1 - 0)^3 + 8 = 64 + 8 = 72.00 \]

Optimistic

Assumes functions without coverage have 100% coverage. This gives the minimum possible CRAP score - useful for surveys where you want to see the "best case" picture.

go-crap scan --missing optimistic

A function with CC=8 and 100% assumed coverage:

\[ CRAP = 8^2 * \left(1 - 1\right)^3 + 8 = 0 + 8 = 8.00 \]

Skip

Excludes functions without coverage from the output entirely. Useful for ignoring untested functions and focusing only on those that have coverage data.

go-crap scan --missing skip

Functions without coverage appear in the output with Skipped: true and their CRAP score equals their complexity.

When to Use Each

• pessimistic - CI checks, code reviews, finding the worst issues first
• optimistic - rough estimation, best-case scenario analysis
• skip - auditing covered code only, ignoring untested functions

Coverage Unavailable vs. Missing Coverage

--missing policy handles functions that have no coverage data because they were not exercised by tests. This is different from coverage unavailable, which occurs when an entire module's go test run fails (build errors, test panics, etc.).

When coverage is unavailable: - go-crap emits a CoverageWarning on affected entries - All formatters surface the error (see scan --help for per-format details) - The --missing policy still applies, but the warning indicates the root cause is a module-level failure, not simply untested functions