[bfs2021]

Problem

subenum -tui exits 0 when the results file could not be written, while the CLI exits 1 in the same situation (Fixes #155). Scripts that launch the TUI and check $? treat a truncated or missing results file as success.

Root cause

reportOutputErr recorded the failure only as a message in the scan view. exitCode consults scanView.failed(), which tracks scan errors alone, so an output-file error never reached the process exit code.

Fix

Track the failure on the model (outputErr) when reportOutputErr is called and return 1 from exitCode when it is set. This mirrors the CLI, where a file error outranks "no changes".

Testing

• go test ./internal/tui/ passes. The new TestOutputFileFailureSetsExitCode drives a real finalizeOutput failure (a directory blocking the final rename), asserts exit code 1 and the on-screen message, and checks that a clean model still exits 0.
• gofmt -l clean.

[TMHSDigital]

Thank you for this. The root-cause write-up was clear and accurate.

#155 was fixed on main in b79b6cd before this PR was reviewed, with the same change (an outputErr flag set by reportOutputErr and checked in exitCode; on main it lives on the scan view, so starting a new scan resets it). This branch now conflicts with main.

Your test is better than the one on main: it makes finalizeOutput really fail instead of injecting an error. I've taken TestOutputFileFailureSetsExitCode into main in 227e756, with you as co-author, plus a CHANGELOG entry. It passes against the code on main.

Closing this as superseded. Thanks again, and contributions are welcome on the other open issues.