v1.2.0

DataLab Version 1.2.0 (2026-04-20)

✨ New Features

Plugin configuration dialog:

DataLab now provides a dedicated plugin configuration dialog (accessible via "Plugins > Configure plugins...") that gives full control over third-party plugin management:

• Enable or disable individual plugins using checkboxes, or toggle all plugins at once with a tri-state master checkbox
• Filter plugins by status: all, enabled, disabled, or plugins with import errors
• View plugin details including version, author, and expandable long descriptions directly in the dialog
• Plugins with import errors are displayed prominently at the top with their full traceback, making it easy to diagnose installation issues
• The expandable text widget used for long descriptions computes its preferred width from a fixed measurement context, ensuring stable layout and reliable "Show full description" toggling regardless of dialog resizing or offscreen rendering

Plugin hot-reload:

• Third-party plugins can now be reloaded at runtime without restarting DataLab, via "Plugins > Reload plugins"
• Enabling or disabling plugins in the Preferences dialog or plugin configuration dialog takes effect immediately - no restart required
• The Plugins menu, status widget, and plugin actions are automatically refreshed after configuration changes

Multi-instance detection:

DataLab now detects when another instance is already running and warns the user before opening a second instance:

• Uses a PID-based lock file mechanism that supports multiple concurrent instances (reference counting)
• Stale PIDs from crashed processes are automatically cleaned up
• Cross-platform support (Windows, macOS, Linux) using platform-specific process detection
• Closing one instance no longer incorrectly removes the lock file when other instances are still running

Image ROI editor contrast synchronization:

• The image ROI editor now shares contrast (LUT) settings with the source image panel
• Adjusting the contrast in the ROI editor is reflected back in the main panel and vice versa
• Contrast controls are fully re-enabled in the image ROI editor dialog

Remote control API — push modified objects back to DataLab:

The proxy API (XML-RPC and Web API) now exposes a new set_object method that updates an existing signal or image in DataLab from a modified copy obtained via get_object (fixes Issue #305):

• Previously, modifications to object properties (e.g. dx, dy, x0, y0, title) made on the result of get_object were lost because get_object returns a copy — set_object now provides a clean round-trip workflow
• Works for both signal and image objects: computed result items attached to the object are preserved during the update, so updating an ImageObj no longer triggers a type mismatch
• The properties panel is automatically refreshed after set_object, so updated object properties (title, units, axes, uncertainties, etc.) are immediately visible in the GUI

📖 Documentation

• Added API documentation for the datalab.objectmodel module
• Added screenshots for the "Paste metadata" dialog (signal and image panels)
• Updated plugin documentation to describe the new configuration dialog, hot-reload workflow, and plugin API helpers
• Updated third-party plugin development guide with new template references and test coverage information
• Expanded Web API reference: documented the binary data transfer options (?compress=false for faster uncompressed NPZ downloads, ?overwrite=true for atomic replacement of existing objects), the in-place PUT /objects/{name} endpoint that updates an object while preserving its identity, group membership and position, and the new "Computation" section listing the /select and /calc endpoints used to drive DataLab computations remotely
• Updated French translations across all new and modified documentation pages

🔧 Improvements

Compatibility:

• Officially support pandas 3.0.x (updated dependency constraint from < 3.0 to < 3.1)
• Updated minimum Sigima requirement from 1.1.0 to 1.1.2 to benefit from latest computation engine fixes and improvements
• Added legacy support for the WINPYDIRBASE environment variable for WinPython-based deployments

Plugin system hardening:

• Plugin discovery and registration is now resilient to third-party import failures while preserving error reporting in the console, logs, and configuration dialog
• Plugin submenus are now scrollable to prevent overflow when many plugin entries are registered

Development tooling:

• New run_with_env.py script for running tasks across multiple Python environment contexts (WinPython, venv, etc.)
• Simplified environment variable handling by removing the DATALAB_ENV_LOADED system
• Fixed Coverage full VS Code task to properly use the run_with_env.py wrapper

🛠️ Bug Fixes

HDF5 workspace - Table serialization:

• Fixed callable metadata not being stripped during HDF5 save/load of TableResult objects
• Fixed string-based enum values in table results not being restored correctly after HDF5 round-trip
• Fixed column_formats attribute not surviving HDF5 round-trip for both TableResult and TableResultBuilder outputs

Plugin system:

• Fixed AttributeError in plugin configuration dialog when clicking "Show full description" (incorrect attribute reference)
• Fixed plugin import errors being silently swallowed when they occurred before the internal console was initialized