PF Simulation r34 — Member-controls maintenance build

Retained member-controls revision. The corrected interface behavior below remains in the current r34 package. This document preserves the original maintenance build’s verification commands and evidence scope. For the current public-service and UI-correction delivery, use python release_tools/package_r34_public_service.py verify --package and read current validation plus the lifecycle upgrade and recovery guide.

This is a minor bug-fix build of r34. It corrects the hosted Team interface without changing the database schema, numerical engines, stored project grants or server authorization rules.

Corrected behavior

• The Organization member selector displays a prompt instead of an unexplained blank field when no member is selected.
• Saving a project grant keeps the selected member and reloads the saved role from the server response.
• Choosing an existing project member displays their saved Viewer or Editor role. A member without a grant defaults to Viewer.
• Organization-role caps remain visible in the role editor. A stale Editor choice cannot persist when the selected member is an organization viewer.
• Project and organization changes clear unrelated selection state. Refreshed member lists discard unavailable selections. Delayed access responses cannot overwrite controls for a different project or logged-out session.

The original blank selector was a display-state defect: the saved grants shown above it remained intact. No repair or recreation of those grants is required.

Upgrade

Use the normal deployment procedure to replace the application files and restart the hosted API service. Reload open browser tabs so they receive the corrected portal script. This minor build adds no migration beyond the existing r34 schema requirements. Existing grants and research-AI policies remain in effect.

Validation and integrity

The build uses focused browser checks of the member editor, actual local-server GUI/API checks, project-authorization tests and release-integrity tests. The exact commands, outcomes, source hashes and retained artifacts are recorded in validation_r34_gui_bugfix/test_execution.json. Its source freeze and per-file review bind these results to the delivered files.

The full 1,822-item regression suite was not rerun. The previous adoption acceptance, including its 454 subtests and separate acceptance groups, remains unaltered under validation_r34_adoption. Those historical results retain their original source identity. Current qualification records explicitly reconcile the reviewed maintenance changes and preserve the original scientific statuses, scopes and limitations; they do not imply new scientific validation.

From the extracted package root, verify source, evidence and every packaged file:

python release_tools/validate_r34_gui_bugfix.py verify --package

MANIFEST.json identifies the current maintenance records. PACKAGE_CHECKSUMS.json covers the packaged files except itself. The downloadable transport has a separate SHA-256 checksum. Changed predecessor files, including its manifest and checksum inventory, are retained in the maintenance evidence capsule; their original hashes and test receipts are not rewritten.

The .zip.xz download contains a ZIP archive. Decompress the XZ layer first, then extract the ZIP. On systems with XZ installed:

xz -dk PF_Simulation_Standalone_and_SaaS_r34_member_controls_fix.zip.xz
unzip PF_Simulation_Standalone_and_SaaS_r34_member_controls_fix.zip

The recorded checks exercise a local hosted application and browser. They do not constitute a new production-host, live external-provider or full scientific regression acceptance run.