Skip to content

Do not retain a previous reconstruction report after a failed run #111

Open

Description

@SamGu-NRX

At PR #20 head d072729, recon/recon/pipeline.py:127-147 writes new model, geometry, scene and coverage files before calling the server. If server.place() raises, execution never reaches the new report write. A reused output directory retains the preceding run's report.md beside the new artifacts and refusal body.

Clear the previous report before a new run can fail, or publish the full output set atomically. Test a reused output directory with an existing report and a mocked server failure, without reconstruction or model jobs. The final directory must not describe a previous reconstruction as the current result.

Confirmed P2 follow-up under the manager's convergence rule. The worker remains an unused draft handoff.

Review: #20 (comment)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions