[Anthony Sligar (sligara7)]

Scope

Follow-up to Nghia's operator reports and Jakub's review, based on merged
#125 (1518646). No QueueServer or process architecture changes.

Delivered

• 6 — GeRM statistics: enable callbacks and statistics computation in
  GeRMStatsDataLogic.prepare_single, for TIFF and HDF data paths. No
  device-specific enabling in EDXD plans.
• 7 — EDXD recovery: restore scan-start motor positions, not requested
  scan-range starts, on success/failure/abort. Grid recovery uses the
  reset_positions_decorator on _body, as requested. Preserve reset opt-outs
  and pause semantics; Halt intentionally bypasses cleanup.
• 8 — Manual shutters: reserve both controls synchronously before deferred
  submission and disable them until completion; guard against busy/paused RE.
  Keep PR125's zero extra manual settling sleep, explicitly approved by AJ;
  acquisition-plan settling is unchanged. Supply device discovery and command
  policy from a HEX adapter rather than the generic widget.

Explicitly deferred

At AJ's direction, remove the device-specific plan-widget special cases.
5 (slit prefill) and 9 (Two theta override) are not delivered here.
Generic forms use plan signatures as before.

11 is now delivered at the plan layer: edxd_scan and
edxd_custom_pos_list_grid default to save_as_tiff=True. The unchanged generic
GUI derives its checked checkbox from that signature. Python callers omitting
the argument also select TIFF; explicit False still selects HDF5.
A later, separate PR may extract the current GUI/config boundary; it will have
nothing to do with QueueServer.

10 — p4p/exporter: unresolved workstation p4p cleanup error. Local
p4p4.3.0 has p4p.server and ordinary shutdown/explicit cleanup exit cleanly.
The installed launcher references the removed gui environment; its deployment
owner and exact workstation state need operator diagnostics. Document supported
exporter task and read-only diagnostics; do not claim a p4p fix.

PR125 cleanup

• Fix CLI example to Pass Through, update constants-format docs and remove
  obsolete filter schema. The separate migration document was removed at AJ's request.
• DCLM/filter restructuring is removed at AJ's request: those source files,
  profile construction and existing related tests match merged PR125 exactly.
  Dataclass/config validation, independent filter settings keys and selected-only
  resolution are deferred to a separate later proposal, not delivered here.
• Pin Pyright, remove unnecessary ignore/conversion, correct misleading profile
  timeout comment without altering device timeouts.
• Do not restore magics, Sample Small display labels, auto-tune plots,
  between-aperture behavior, or compatibility aliases removed by PR125.

Verification

Own worktree's locked Pixi environment, with CA/PVA automatic discovery disabled
and addresses restricted to loopback. No borrowed environment/PYTHONPATH override.

• Prior expanded review suite: 263 passed (saved execution log).
• Final reviewer-fix regression run:
  pytest tests/test_review_followup.py tests/test_edxd_operator_recovery.py tests/gui/test_generic_widget_boundary.py tests/gui/test_shutter_status.py:
  45 passed.
• pixi run --locked lint: passed.
• pixi run --locked type-check: Pyright, 0 errors.
• Actually reran pixi run --locked hex-gui-mock: inspected generic slit/theta
  defaults, exercised both shutter actions and mutual exclusion, captured
  screenshots, and exited the GUI. No removed special-case behavior claimed.
• TIFF follow-up: 54 tests passed, including actual emitted TIFF resources
  when the argument is omitted, explicit True/False formats, EDXD recovery and
  generic checkbox/boundary tests. Pyright/lint passed; actual mock GUI rerun
  confirms checked TIFF and user opt-out, with no GUI production-code change.
• After removing DCLM/filter restructuring: 85 tests passed, covering restored
  DCLM/filter/stop tests plus retained GeRM, EDXD recovery, shutter and GUI-default
  regressions. Configured Pyright and lint passed.

The boundary tests cover the touched generic widgets, not complete architectural
separation of the entire current GUI. Known device-specific policy is not added
to generic widget code.

Mock-only verification: no beamline hosts, live PVs, IOC restart or hardware
validation. No Docker resources used or pruned.

[Copilot]

🟢 Approval recommended

The review reports no active issues requiring resolution.

No open issues 🎉

What changed in this PR

Addresses HEX operator issues in slit controls, EDXD acquisition recovery, and manual shutter operation, with documentation for launcher troubleshooting.

Changes:

• Loads selected slit values asynchronously, rejects stale responses, and labels the tilt input Two theta.
• Re-enables GeRM statistics before acquisition, restores pre-scan positions, and prevents overlapping manual shutter actions.
• Adds regression coverage and documents recovery behavior, supported exporter launch commands, and workstation diagnostics.

File	Description
tests/​test_edxd.py	Corrects a comment describing position restoration.
tests/​test_edxd_operator_recovery.py	Tests statistics recovery, position restoration, interruptions, and reset opt-outs.
tests/​gui/​test_shutter_status.py	Tests shutter exclusion and busy/paused RunEngine gating.
tests/​gui/​test_plan_widget_slits.py	Tests slit readbacks, asynchronous loading, stale responses, and errors.
tests/​gui/​test_edxd_tilt_label.py	Verifies the tilt label preserves the submitted theta argument.
src/​hextools/​gui/​shutter_status.py	Reserves shutter controls during deferred execution and gates them on RunEngine state.
src/​hextools/​gui/​plan_widget.py	Adds asynchronous slit-value loading and the tilt label override.
src/​hextools/​gui/​device_sources.py	Adds local device-path resolution for slit reads.
src/​hextools/​edxd.py	Enables GeRM statistics and restores initial positions in 1D/grid scans.
docs/​gui.md	Documents slit loading, shutter exclusion, and EDXD recovery behavior.
docs/​getting_started.md	Documents exporter launching and local launcher/p4p diagnostics.

_{Review effort: Balanced}

[Anthony Sligar (sligara7)]

Integration update — 2026-10-29

This draft now incorporates #125 at
b2f5550f5b10ab76cd163f9b02593d1e72e0e806, as requested by AJ.
Do not merge until PR125 has merged. Until then, the comparison to main
also includes PR125's changes. PR125 itself is untouched; no published history
was rewritten.

• Item11: EDXD scan/custom-position grid GUI Save as TIFF starts checked and
  remains toggleable. Python plan defaults are unchanged.
• AJ explicitly approved PR125's zero extra manual-button settling sleep,
  combined with our mutual-exclusion guard. Acquisition-plan timing is unchanged.
• PR125 switches the configured type checker to Pyright; corrected the slit
  field types accordingly.

Verification after integration:

• Expanded suite: 221 passed, one newly added TIFF test initially failed due
  to a missing mandatory positions-list fixture. Corrected that fixture.
• Final focused slit/TIFF/shutter tests: 12 passed.
• Configured lint and Pyright: passed.
• Actual pixi run --locked hex-gui-mock rerun: slit and tilt submissions,
  shutter exclusion and visibly checked TIFF checkbox verified; GUI exited.
• Local p4p4.3.0 has p4p.server; ordinary shutdown and explicit cleanup each
  exited0 with empty stderr. The reported workstation p4p error remains
  unresolved, not fixed by this PR. No live host/PV access.

This update supersedes the original PR body's main-only base and retained
manual-button settling-delay statements below.

[Anthony Sligar (sligara7)]

At AJ’s request, removed docs/migration-pr125.md and its docs index entry. Clarification: merged PR125 defines set_fe_aperture, and this PR does not change that name or add the old name back. Jakub’s merged API remains authoritative.

[Anthony Sligar (sligara7)]

At AJ’s request, removed the DCLM/filter restructuring and coupled profile/test changes. These files now match merged PR125 exactly. Kept GeRM lifecycle statistics enabling, EDXD recovery, manual shutter exclusion and plan-level TIFF defaults. Restoration verification:85 tests passed; configured Pyright/lint passed. Restructuring is deferred to a separate later proposal.

[Copilot]

🟡 Changes recommended

The GUI tests import optional dependencies unconditionally, causing pytest collection to fail instead of skipping them when GUI extras are absent.

2 open issues

• Guard optional GUI imports during test collection · New
• Guard optional Qt dependency to skip GUI test · New

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

The changes control live beamline hardware, but verification is mock-only, requiring a beamline operator to confirm safe motor recovery and shutter-completion timing.

No open issues 🎉

2 resolved since last review

• Guard optional Qt dependency to skip GUI test
• Guard optional GUI imports during test collection

_{Review effort: Balanced}

[Anthony Sligar (sligara7)]

Checked compatibility with Jakub’s PR127 at720210f. The only merge conflict was overlapping CLI filter documentation; ebafcd6 adopts PR127’s exact wording/example. git merge-tree now succeeds without conflicts. Both PRs delete the obsolete filter schema consistently. PR126 does not modify DCLM or its tests, so PR127’s DCLMConstants/.constants implementation is preserved. No PR127 code was merged into this branch; this is a source/merge compatibility check, not combined runtime validation.

[Copilot]

🔵 Needs a closer look

Human assessment is needed because a critical cleanup-ordering risk can leave shutters open during motor recovery, prolonging sample exposure after success, failure, or abort.

No open issues 🎉

_{Review effort: Balanced}