v0.1.0b1 (beta-nightly)

First beta release. Core GUI, all 15 analysis modules, batch processing, NWB export, and plugin interface are functional. Issued as a pre-release for wider testing before a stable 0.1.0 tag.

Added

• Sag Ratio Standalone Analysis: Promoted sag_ratio_analysis from an inline
  computation within rin_analysis to its own registered analysis plugin (15th
  built-in module). Provides dedicated UI controls for baseline, peak, and
  steady-state windows, Savitzky–Golay smoothing, and rebound depolarisation
  measurement. Results: sag_ratio, sag_percentage, v_peak, v_ss,
  v_baseline, rebound_depolarization.
• Documentation: Updated tutorial (§4.15), algorithmic definitions (§4),
  API reference, user guide, developer guide, README, and extending guide to
  reflect the new standalone sag ratio analysis and 15 built-in modules.
• Deferred Initial Reset for Multichannel: Generation-counter-protected
  _deferred_initial_reset() catches post-layout sigResized shifts for
  multichannel recordings without interfering with view state restoration.
• Custom Analysis Plugin Documentation: Comprehensive guide (extending_synaptipy.md) for writing analysis plugins without modifying source code
• Plugin Template: Ready-to-copy template at src/Synaptipy/templates/plugin_template.py with inline comments for all parameter types and plot overlays
• Plugin Tests: 16 tests validating plugin template logic, PluginManager loading, and wrapper conventions
• Tutorial Section 3.6: Step-by-step "Adding Your Own Analysis Tab" section in the user tutorial under the Analyser Tab
• Stress Tests: File cycling stress tests for plot canvas rebuild stability (100 iterations)
• Explorer Debounce: Debounce timer for file navigation to prevent rapid teardown cycles
• Regression tests for registry population (test_registry_metadata.py)
• Regression tests for preprocessing reset propagation (test_preprocessing_reset.py)
• Developer documentation for registry import rule, editable install, and
  preprocessing reset propagation (copilot-instructions.md, developer_guide.md)

Fixed

• Explorer X-Axis Shift on File Cycling: Fixed bug where the X-axis would
  shift right (not starting at 0) when cycling through files, especially with
  multichannel recordings. Root cause: old ViewBoxes scheduled for
  deleteLater() continued to emit sigXRangeChanged / sigYRangeChanged /
  sigResized signals after the widget was replaced, corrupting slider and
  scrollbar values for the new recording. Fix: explicitly disconnect all
  ViewBox signals in ExplorerPlotCanvas.rebuild_plots() before clearing plot
  items.
• Explorer X-Link Range Corruption: Fixed multichannel X-range corruption
  caused by linkedViewChanged() recalculating ranges from screen-geometry
  pixel offsets between stacked ViewBoxes. _reset_view() now blocks link
  propagation via ViewBox.blockLink(True) while setting ranges, then unblocks.
• Overlay Mode Y-Range Too Narrow: Fixed _compute_channel_y_range() to
  compute Y range from all trials (sampling up to 50 evenly spaced) instead of
  only trial 0. Previously, if trial 0 was at resting potential but other
  trials contained action potentials, the Y range was too narrow to display
  the full signal.
• Flaky Qt Tests: Added processEvents() calls in 3 test files to resolve non-deterministic offscreen failures caused by stale deferred ViewBox geometry callbacks
• Explorer Plot Layout: Fixed Windows Explorer plot view state preservation during file cycling
• Lint Errors: Resolved all flake8 CI failures in analysis_formatter and exporter_tab
• CSV Export: Updated tidy per-type CSV export for batch results
• Windows Analysis Loading: Fixed registry import bug where AnalysisRegistry remained
  empty on Windows because only registry.py was imported (not the full
  Synaptipy.core.analysis package that triggers @register decorators). Added
  import Synaptipy.core.analysis in analyser_tab.py and startup_manager.py.
• Preprocessing Reset: Connected the preprocessing_reset_requested signal in
  BaseAnalysisTab and added _handle_preprocessing_reset() handler. Added
  reset_ui() method to PreprocessingWidget. Reset now propagates globally to
  all sibling analysis tabs via AnalyserTab.set_global_preprocessing(None).

Analysis Module Bug Fixes

• Tau (Time Constant): Added exponential fit overlay plot — calculate_tau now returns
  fit curve data (fit_time, fit_values) alongside tau_ms, and the registration
  includes overlay_fit plot metadata so the fit curve is drawn on the main trace
• Excitability (F-I Curve): Added popup_xy plot metadata to show F-I Curve popup
  (Frequency vs Current) after multi-trial analysis
• Spike Train Dynamics: Added ISI popup plot — wrapper now returns isi_numbers and
  isi_ms arrays, and registration includes popup_xy plot metadata
• Optogenetic Synchronization: Added secondary channel selector (requires_secondary_channel
  metadata) so users can pick a dedicated TTL/trigger channel instead of falling back to
  the voltage trace; added stimulus onset vertical line markers to the plot
• Event Detection (Template Match): Lowered default threshold from 4.0 SD to 3.0 SD
  for better sensitivity; added direction parameter to UI so users can switch polarity;
  fixed time-axis reconstruction to use actual time array instead of synthesising from
  sampling rate (fixes event time accuracy when data doesn't start at t=0)
• Event Detection (Threshold): Fixed noise-floor guard that could override user threshold —
  the 2-SD noise guard now only activates when the user's threshold is below 1 SD, otherwise
  the user's explicit threshold is honoured
• Added overlay_fit visualisation type to MetadataDrivenAnalysisTab for drawing
  analysis fit curves on the main plot
• Added _inject_secondary_channel_data to MetadataDrivenAnalysisTab for loading
  data from a user-selected secondary channel and passing it to analysis functions

Publication Readiness Audit — Error Handling & Robustness

• Replaced ~25 silent except: pass blocks with diagnostic log.debug() calls across 15 files
• Added logging to error-swallowing blocks in neo_adapter, analysis_formatter, explorer_tab,
  plot_canvas (widgets & explorer), analysis_plot_manager, main_window, startup_manager,
  zoom_theme, plot_customization, and analysis_tabs/base
• Added docstrings to NWB exporter fallback sentinel classes
• Removed trailing pass statements and dead placeholder code from explorer_tab,
  shortcut_manager, main_window, file_io_controller, data_loader, and base analysis tab
• Cleaned up duplicate comment in base analysis tab error handler
• Removed unnecessary pass after log.debug in theme_manager

Publication Readiness Audit — Code Quality

• Fixed unused variable param_key in metadata_driven analysis tab
• Fixed unused variable dt in optogenetics wrapper
• Replaced long conditional expressions with readable intermediate variables in
  optogenetics.py and train_dynamics.py
• Rewrote conversational docstring in optogenetics wrapper with proper Args/Returns format
• Fixed all flake8 violations: trailing whitespace, missing blank lines, line length, W391
• Cleaned up CLI placeholder module with proper docstrings (removed 50 lines of dead scaffolding)
• Removed stale CSVExporter comments from exporters __init__.py

Publication Readiness Audit — Package Structure

• Populated __all__ exports in application/controllers/__init__.py (9 symbols)
• Populated __all__ exports in application/gui/analysis_tabs/__init__.py (3 symbols)
• Added __all__ and module docstring to application/gui/explorer/__init__.py

Publication Readiness Audit — CI/CD

• Added Python 3.12 to CI test matrix (now tests 3.10, 3.11, 3.12)
• Added pytest-cov coverage reporting to CI (--cov=Synaptipy --cov-report=term-missing)
• Added coverage XML artifact upload for ubuntu/3.11 builds

Publication Readiness Audit — Scientific Accuracy

• Fixed line-noise detection baseline overlap in signal_processor.py
• Fixed max/min dV/dt zeroing bias in spike_analysis.py (sentinel values)
• Fixed AHP depth sign convention in spike_analysis.py
• Fixed mean frequency calculation to use spike span instead of trace duration
• Fixed Rin unit conversion clarity (mV/pA → MOhm derivation)
• Refined sag ratio calculation to use 5th percentile for robustness
• Fixed z-score normalization in template matching (subtract median)
• Added dV/dt unit conversion documentation in phase_plane.py
• Removed duplicate dictionary keys in spike detection registry

Publication Readiness Audit — Code Quality

• Removed redundant imports, unused variables, and duplicate function definitions
• Added edge-case handling for empty spike indices
• Added docstring for _find_stable_baseline_segment
• Converted unresolvable TODO to NOTE (async limitation in batch load)
• Standardized all flake8 compliance to max-line-length 120

Publication Readiness Audit — CI/CD & Infrastructure

• Made flake8 lint failures blocking in CI (removed --exit-zero)
• Added pytest-cov to CI dependencies
• Aligned Python version floor to 3.10 in pyproject.toml, environment.yml, README
• Standardized author email and license (AGPL-3.0-or-later)
• Cleaned stale files, relocated tests, removed empty directories
• Added .coverage, htmlcov/, .pytest_cache/ to .gitignore
• Removed stale Python 3.9 classifier from pyproject.toml
• Added pytest-cov and flake8 to environment.yml
• Added 7 targeted scientific accuracy tests

Phase 1: Critical Performance & Data Loading

• CRITICAL: Fixed severe UI lag caused by repeated instantiation of PlotCustomizationManager in get_plot_pens(). Now uses singleton pattern, eliminating thousands of unnecessary disk reads.
• CRITICAL: Optimized `update_pl...