v1.0.3

DataLab Version 1.0.3

🛠️ Bug Fixes since version 1.0.2

Loading images with a single row or column fails:

• Fixed loading image files with a single row or column (e.g., some SIF spectroscopy files with shape (1, N)) causing an IndexError and preventing the image from being displayed
• The underlying issue was in PlotPy's coordinate-to-bin conversion function which assumed at least 2 points
• This fix is included in PlotPy 2.8.3

LUT range incorrectly copied when processing images:

• Fixed processed images inheriting the LUT range (display contrast settings) from the source image, causing incorrect visualization when data values change significantly
• When applying operations like offset correction, the result image now auto-scales its display range to fit the new data values instead of using the original image's LUT range
• Previously, users had to manually adjust the LUT to see processed images correctly (they could appear completely black or with wrong contrast)
• This closes Issue #288 - LUT range incorrectly copied when processing images
• This fix is included in Sigima 1.0.4

Separate plot windows - Incorrect aspect ratio for non-uniform coordinates:

• Fixed separate plot windows (e.g., "View in a new window", ROI editors, profile dialogs) not using the same aspect ratio configuration as the integrated plot handler
• When displaying images with non-uniform coordinates, the main plot correctly unlocks the aspect ratio for proper display, but separate windows were not inheriting this setting
• All image dialogs now properly inherit the aspect ratio settings, ensuring consistent display behavior across all plot windows
• This closes Issue #287 - Separate plot windows don't use same aspect ratio configuration as integrated plot

Remove all results - AttributeError when ROI was removed:

• Fixed "Remove all results" action failing with AttributeError: 'NoneType' object has no attribute 'get_single_roi_title' when results contain ROI information but the ROI was subsequently removed from the object
• The issue occurred when:
  1. Running an analysis with a ROI (e.g., Centroid on an image with rectangular ROI)
  2. Removing the ROI from the object
  3. Trying to delete all results via "Analysis > Remove results > Remove all results"
• The fix adds a proper None check before accessing the ROI's title
• This closes Issue #286 - "Remove all results" fails with AttributeError when ROI was removed

Analysis auto-recompute - Stale parameters after deleting results:

• Fixed analysis parameters not being cleared when deleting analysis results, which could cause unexpected auto-recompute behavior when modifying ROIs
• After deleting results (via "Delete all results" or individual result deletion), changing the ROI would trigger recomputation of the deleted analysis because the stored parameters remained in object metadata
• The fix ensures analysis parameters are properly cleared alongside the results, preventing unwanted automatic recomputation
• This closes Issue #285 - Analysis parameters not cleared after deleting results

Backwards-drawn rectangular ROI causes NaN statistics:

• Fixed rectangular ROI statistics returning NaN values when the ROI was drawn "backwards" (from bottom-right to top-left instead of top-left to bottom-right)
• The issue occurred because the ROI coordinate conversion produced negative width/height values, causing the mask generation to fail
• ROI mask generation now works correctly regardless of the direction in which the rectangle was drawn
• This closes Issue #284 - Backwards-drawn rectangular ROI causes NaN statistics
• This fix is included in Sigima 1.0.4

Plot refresh - ROIs and annotations persist after "Remove all":

• Fixed ROIs and annotations occasionally remaining visible after executing "Remove all" action from the Edit menu
• The issue occurred because the plot widget wasn't explicitly told to redraw after clearing all plot items and shape items
• The plot now correctly refreshes its display after clearing all objects, ensuring a clean view without requiring panel switching
• The bug was intermittent and hard to reproduce systematically, making it appear as if objects were "stuck" on the plot

Result labels - False "omitted" message for single-row results:

• Fixed result labels (e.g., pulse features) incorrectly showing "X more rows omitted" message when displaying single-row results with no actual truncation
• The bug occurred because len(adapter.result) was used to get the row count, but this actually returns the number of column headers, not data rows
• For pulse features with 13 columns, this caused the label to think there were 13 rows when only 1 row existed, leading to a false "12 more rows omitted" message
• The fix ensures accurate row counting using len(df) instead, preventing false truncation notices
• This closes Issue #281 - Result Labels False "Omitted" Message

Signal view - XRangeSelection items persist after removing signals:

• Fixed XRangeSelection and DataInfoLabel items (created by PlotPy's curve statistics tools) remaining visible on the plot after removing all signals
• When activating the XRangeSelection tool from the PlotPy toolbar and then removing the signal(s), these tool-created graphical objects were not cleaned up
• The cleanup logic now correctly triggers when there are zero selected objects (after deleting the last one), not just when exactly one object is selected
• This closes Issue #280 - XRangeSelection items persist after removing all signals

Remote control - Crash when adding object via proxy:

• Fixed AttributeError: 'NoneType' object has no attribute 'setText' crash when adding objects via RemoteProxy.add_object() in macros
• The crash occurred in update_tree() when the tree view became temporarily out of sync with the object model (e.g., when adding objects with specific metadata configurations or ROIs via remote proxy)
• The tree view now rebuilds automatically if inconsistencies are detected, ensuring robust object addition in all scenarios
• This closes Issue #279 - AttributeError in update_tree() when adding object via RemoteProxy

Icons - Qt SVG warnings on Linux:

• Fixed "qt.svg: Invalid path data; path truncated" warnings appearing 4 times at startup on Linux/Ubuntu
• Cleaned SVG icon files to remove Inkscape/Sodipodi metadata and fixed path syntax in two icon files that used implicit cubic Bézier command continuation (not supported by Qt's SVG renderer on Linux)
• This closes Issue #278 - Fix "qt.svg: Invalid path data; path truncated" warnings on Linux

ROI extraction - KeyError when extracting ROIs:

• Fixed KeyError when extracting ROIs from images with ROIs generated by blob detection or other analysis functions
• The issue occurred because the ROI editor was trying to access plot items that hadn't been created yet (e.g., when "auto refresh" was disabled or in certain selection scenarios)
• The ROI extraction dialog now handles missing plot items gracefully
• This closes Issue #276 - KeyError when extracting ROIs from images with blob detection results

Plugin development - Infinite loop when iterating over group objects:

• Fixed infinite loop occurring in plugins when iterating over object IDs returned by ObjectGroup.get_object_ids() while simultaneously adding new objects to the same group
• The method was returning a direct reference to the internal list instead of a copy, causing the iteration to include newly added objects and loop indefinitely
• This particularly affected plugin workflows that loop over signals/images in a group and add results to the same or another group
• The fix ensures get_object_ids() now returns a copy of the list, making iteration safe even when modifying the group
• This closes Issue #274 - Infinite loop when iterating over group objects while adding new objects

Marker visibility:

• Fixed cross marker (shown when pressing Alt key on plot) colors inherited from PlotPy being too pale on white background, and other marker properties explicitly configured for better contrast
• Fixed default annotation colors inherited from PlotPy being too pale on white background - segment and drag shape annotations now use brighter green (#00ff55) for better visibility

Macro panel layout:

• Fixed macro console taking excessive vertical space on first open - the script editor now properly gets 70% of space and console 30% by default, ensuring comfortable editing without manual resizing

Macro naming:

• Fixed automatic macro naming generating duplicate names after reloading workspace from HDF5 files
• When macros were deserialized from saved workspaces or imported from files, the internal counter used for generating default names (e.g., "macro_01", "macro_02") was not updated, causing the next new macro to potentially reuse an existing name
• The counter now synchronizes with existing macro names after loading, ensuring unique sequential naming
• Default macro names simplified from translated "Untitled XX" to language-neutral "macro_XX" format for better consistency across locales

Macro execution:

• Fixed syntax errors when using f-strings with nested quotes in macros (e.g., f'text {func("arg")}' now works correctly)
• Fixed corrupted Unicode characters in macro console output on Windows - special characters like ✅, 💡, and → now display correctly instead of showing garbled text

Internal console - HDF5 workspace operations:

• Fixed application freeze when calling open_h5_files() from the internal console - the console runs in a separate thread and creating Qt GUI elements (progress dialogs) from non-main threads causes deadlocks
• Added new headless API methods load_h5_workspace() and save_h5_workspace() that can be safely called from the internal console or any script context without GUI dependencies
• These methods are also available through the remote control API for consistency
• Note: load_h5_workspace() only supports native DataLab HDF5 files; for importing arbitrary HDF5 files, use macros with RemoteProxy
• This closes Issue #275 - Console freezes when calling open_h5_files() from internal console

Signal cursor dialogs - Decimal input with regional settings:

• Fixed decimal value input failing in signal cursor dialogs when using regional settings with comma as decimal separator (e.g., French, German locales)
• The Y input field in "First abscissa at y=...", "Ordinate at x=...", and "Full width at y=..." dialogs now consistently accepts dot as decimal separator regardless of system locale
• This closes Issue #276 - Decimal input fails with regional settings using comma as decimal separator

Radial profile title:

• Fixed duplicate suffix in result image title when extracting radial profile from an image (e.g., radial_profile(i019)|center=(192.500, 192.500)|center=(192.500, 192.500) instead of radial_profile(i019)|center=(192.500, 192.500))
• This fix is provided by Sigima 1.0.4

Grid ROI - Missing spacing parameters for non-uniform grids:

• Fixed grid ROI feature not working correctly for images where subimages don't fill the entire image area (e.g., laser spot arrays with gaps between spots)
• Added missing xstep and ystep parameters to control horizontal and vertical spacing between ROI centers, as a percentage of the automatically computed cell width/height (default 100% = evenly distributed grid)
• Previously, the grid was always assumed to be evenly distributed across the entire image, which failed when there was significant offset or gaps between features
• Users can now adjust ROI spacing independently from ROI size to accurately extract grids from real-world data like laser spot arrays, diffraction patterns, or any regularly spaced features with gaps
• This fix is provided by Sigima 1.0.4
• This closes Issue #282 - Grid ROI Missing Spacing Parameters

Analysis auto-recompute - Redundant O(n²) calculations when adding ROI:

• Fixed severe performance issue where adding ROI to N selected images with analysis results (e.g., Statistics, Centroid) triggered N² calculations instead of N
• When 10 images were selected with pre-computed statistics, adding a ROI would show 10 progress bars of 10 calculations each (100 total) instead of a single progress bar with 10 calculations
• The issue occurred because auto_recompute_analysis() was calling compute_1_to_0() which by default processes all selected objects, not just the specific object being recomputed
• Added target_objs parameter to compute_1_to_0() to allow specifying which objects to process, and updated auto_recompute_analysis() to use it
• This closes Issue #283 - Redundant O(n²) calculations when adding ROI to multiple images with analysis results

Result visualization - Analysis result segments hard to see:

• Fixed analysis result markers (FWHM, pulse features, etc.) being difficult or impossible to see on signal plots
• These measurement indicators were appearing black on white backgrounds or white on dark backgrounds, making them blend into the plot
• They now display in bright green with thicker lines, making them clearly visible in all color themes
• Also removed the filled area that was obscuring the underlying signal data - markers now show only the boundary lines indicating the measured region