fix: ensure backend is initialized to prevent first plot rendering empty #359
Conversation
Issue #355: First plot appeared empty on GitHub Pages due to backend not being initialized during figure() call, causing plot_count to reset when ensure_global_figure_initialized() was called later. Changes: - Initialize default PNG backend in figure_t%initialize() if not specified - Add default color initialization to plot_data_t type (both versions) - Add test to verify first plot renders with correct colors The bug occurred because: 1. figure() called fig%initialize() without backend parameter 2. Backend remained unallocated 3. Later calls (plot, title, etc.) checked if backend was allocated 4. Since it wasn't, they called initialize() again, resetting plot_count to 0 5. This caused the first plot data to be lost Now the backend is always initialized, preventing the unwanted reset. Fixes #355
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
…e violation - Extract rendering pipeline functions to fortplot_rendering.f90 (511 lines) - Extract contour algorithms to fortplot_contour_algorithms.f90 (144 lines) - Reduce fortplot_figure_core.f90 to 866 lines (from 1776 lines) - Preserve Issue #355 fix throughout refactoring - All modules now comply with QADS hard limit of 1000 lines per file - Maintain all existing functionality and test compatibility
Refactoring Complete ✅The CRITICAL file size violation has been resolved through module refactoring: Before:
After:
Total: 1521 lines across 3 well-organized modules All tests pass and Issue #355 fix is preserved. The PR is now ready for review. |
Patrick's Code Quality Review - APPROVED ✅Critical Issue Resolution✅ File size violation RESOLVED:
✅ Issue #355 fix preserved: Backend initialization maintained, test passes ✅ Build issues fixed: All import errors resolved during review Refactoring Quality
Non-Critical FindingFiled Issue #360: ConclusionExcellent refactoring that resolves the critical blocking issue while preserving functionality. Ready to merge. HANDBACK TO MAX: All critical issues resolved. PR ready for final merge. |
Summary
Fixes Issue #355 where the first plot appeared empty on GitHub Pages. The issue was caused by the backend not being initialized during the
figure()call, which led to plot_count being reset whenensure_global_figure_initialized()was called later.Problem
The bug occurred because:
figure()calledfig%initialize()without backend parameterplot,title, etc.) checked if backend was allocatedinitialize()again, resetting plot_count to 0Solution
figure_t%initialize()if not specifiedplot_data_ttype (both versions)Test
Added
test_first_plot_rendering.f90which verifies that the first plot renders with proper colors (not just dots).Fixes #355
🤖 Generated with Claude Code
Co-Authored-By: Claude noreply@anthropic.com