Save Focus doesn't work in Protocol

[EtalumaSupport]

I made a simple 96 well BF Protocol and the Saved Focus was not used when New was clicked.

SN12064-2026-05-28-070736.zip

[EtalumaSupport]

Seems to go to 7000 regardless of Saved Focus in the channel

[EtalumaSupport]

Clicking Save Focus then Goto Focus more than once seems to fix it.

[ericweiner]

Beta14 fix proposal (LVP color-audit-phase1 @ 53704df)

Root cause: Save Focus only wrote settings[layer]['focus']. The in-memory protocol's per-step Z stayed at whatever was loaded (e.g. 7000 from data/new_default_protocol.tsv). At the next New Protocol click, the previous-well-Z carry-over from #535 won over layer_config['focus'], so the newly-saved focus was silently ignored. Hence "Saved Focus was not used when New was clicked."

Fix: Protocol.update_layer_focus(layer, old_z, new_z) walks the in-memory step table and updates Z on every step whose Color matches the layer AND whose current Z equals old_z (within 1e-3 um). LayerControl.execute_save_focus reads the old baseline before writing the new value, then calls the helper on ctx.protocol. Per-well-tuned steps (Z != old_z) are preserved untouched, so #535's intent (per-well-tuned Z survives New) co-exists with the fix.

Cluster check: Single-site fix. Sibling layer settings (illumination / gain / exposure) are also stored per-step but the user-mental-model treats them as live-preview adjustments, not as "saved" anchors — no propagation needed for those.

Regression test: tests/test_issue_681_save_focus_propagates_to_protocol.py -- 7 tests including the at-baseline-update path, the user-tuned-preserve path, other-layer-isolation, no-baseline-on-first-save, empty-protocol, float-tolerance, and an AST structural lock on execute_save_focus. The #535 suite still passes alongside (4 tests in test_issue_535_z_carry_over_on_new.py).

Test plan to confirm fix:

1. Load new_default_protocol.tsv (or any protocol where every well shares the same default Z).
2. Move Z to a chosen focus position; click Save Focus on BF.
3. Click New Protocol.
4. Click Scan — stage should move to the saved focus on the FIRST scan attempt. No more click-Save-Goto-multiple-times workaround.

Status: code-side resolved on color-audit-phase1; cherry-pick to 4.0.0-beta after bench validation. Hardware verification still pending -- please test before closing.

[ericweiner]

Now on 4.0.0-beta (via the color-audit-phase1 merge)

The Save Focus fix from 2026-05-28 (53704df, plus the per-step focus-Z readout cue 2ea9092) is now on the shipping 4.0.0-beta lineage -- both commits are confirmed ancestors of the current beta HEAD, so no separate cherry-pick is needed. The prior annotation described this as "resolved on color-audit-phase1, cherry-pick pending"; that branch has since been absorbed into beta.

Ready to verify on a beta build. Retest per the test plan in the earlier comment: load a protocol where every well shares the default Z, click Save Focus on BF, click New Protocol, then Scan -- the stage should move to the saved focus on the first attempt (no Save-then-Goto-repeat workaround). The per-well-tuned-Z behavior from #535 is preserved.

Status: fixed on 4.0.0-beta; hardware verification pending -- please test before closing.

[EtalumaSupport]

New Blank with 3 channels, each at a different saved focus, 7000, 8000, 9000 in this case. Add results in three Steps all with 8000 focus level. It appears that all Steps get the level of the last Saved Focus before Adding.

SN12064-2026-06-01-070018.zip

[EtalumaSupport]

Seems to only happen with Blank labware, 96 well Protocol works.