[Jakub Wlodek (jwlodek)]

• Fixes to germ exporter, improve auto exporter, remove auto exporter from gui for now
• Fix issue where scan browser with large search result would resize window height

[Copilot]

🟡 Changes recommended

The scrolling fix lacks a regression test verifying that overflowing results enable the table’s vertical scrollbar without increasing the window height.

1 open issue

• Add regression test for scan-browser scrolling in tab widget · New

What changed in this PR

This PR addresses scan-browser window growth when searches return many results and moves the browser into its own top-level GUI tab.

Changes:

• Adjusts table sizing and scrolling policies to keep large result sets within the available space.
• Displays search errors on one line, with full details in a tooltip.
• Makes the Scan Browser a separate tab and removes the local RunEngine requirement for creating it.

File	Description
src/​hextools/​gui/​scan_browser.py	Updates table sizing, scrolling, and error display to prevent window growth.
src/​hextools/​gui/​__main__.py	Moves Scan Browser to a top-level tab and broadens its availability.

_{Review effort: Balanced}

[Copilot]

🟡 Changes recommended

Metadata retrieval failures can silently broaden searches across all accessible proposals, and the outdated browser tab selection breaks screenshot generation.

2 open issues

• Select the scan browser through the outer tabs in screenshot script · New
• Propagate metadata errors to avoid dropping the proposal filter · New

1 resolved since last review

• Add regression test for scan-browser scrolling in tab widget

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

QueueServer users cannot access the new metadata-backed search because scan-browser creation is guarded by a plan monitor that is only constructed for a local RunEngine.

2 open issues

• Propagate metadata errors to avoid dropping the proposal filter
• Select the scan browser through the outer tabs in screenshot script

1 previously missed issue

In code that hasn't changed since last review

Create scan browser for QueueServer clients

src/​hextools/​gui/​__main__.py:572

The new QueueServer support is unreachable here: this block is inside if self._plan_monitor is not None, but the monitor is only constructed for a local RunEngine. With a RunEngineClient, scan_browser remains None even when a Tiled client is available. Move browser creation outside the plan-monitor guard so QueueServer users can access the new metadata-backed search.

_{Review effort: Balanced}

[Copilot]

🟡 Changes recommended

Searches using RunEngineClient fail with AttributeError because the locked bluesky-queueserver-api 0.0.13 does not provide re_metadata().

1 open issue

• Allow a QueueServer API version that supports re_metadata() · New

2 resolved since last review

• Propagate metadata errors to avoid dropping the proposal filter
• Select the scan browser through the outer tabs in screenshot script

_{Review effort: Balanced}

[Copilot]

🟡 Changes recommended

The plan-name change breaks alignment spec validation and its test, while appending six-column rows under an existing five-column CSV header silently corrupts column mappings.

2 open issues

• Appending new rows corrupts existing CSV schema · New
• Plan name mismatch causes alignment spec validation to fail · New

1 resolved since last review

• Allow a QueueServer API version that supports re_metadata()

_{Review effort: Balanced}

[Anthony Sligar (sligara7)]

Reviewed and looks good. A few nits:

• Default CSV location (low).
  • When the proposal directory is missing, as in CI, the write fails and is logged at error level with a traceback for every run. That's noisy, but nothing breaks.
  • If someone passes an existing --csv file, the new rows have a different column order from its old header.
• Search scope (low, already known). With no data_session, the search quietly covers every proposal the user can see. Copilot flagged this; the old code did the same.

[Copilot]

🟡 Changes recommended

The plan-name change invalidates historical alignment runs without a spec-version bump, while appending six-field CSV rows under existing five-column headers silently corrupts column mappings.

3 open issues

• Plan name mismatch causes alignment spec validation to fail
• Appending new rows corrupts existing CSV schema
• Preserve validation for historical alignment plan names · New

_{Review effort: Balanced}