Debugging and diagnostics

Begin at the boundary where the failure appears: target discovery, transaction preview, native command, runtime startup, or a particular plugin service.

Read-only target diagnostics

cargo run -p fluxplugs-injector-core --bin diagnose-targets -- /path/to/Fluxer

Use a quoted installation root on Windows. The command emits JSON diagnostics without injecting or restoring the installation.

In the desktop application, Check status refreshes the selected installation. Process state and backup health are different observations; an unknown process state is not proof that Fluxer is stopped.

Reproduce safely

Use fake-Fluxer to investigate mutation behavior. Keep the archive’s before/after hashes and the operation preview. For a stale-preview failure, inspect what changed between review and commit.

For runtime issues, check API compatibility, manifest identity, code hash, grants, and startup errors before tracing RPC. A blocked operation may be correctly enforcing policy.

Logging and issue reports

Provide versions, operating system, reproducible steps, expected behavior, actual behavior, and relevant sanitized diagnostics. Omit credentials, private message content, personal file paths, and signed attachment URLs.

Report a bounded failure rather than attaching an entire application-data directory. Add a regression test to the owning package once the cause is understood.