Fix interface out of bounds error on MSYS2

[rory-cd]

Description

Problem

Interfaces have a bug which can cause out-of-bounds errors in MSYS2. When running the UI test in MSYS2 via sktest.exe, the following error is received:

C:/msys64/mingw64/include/c++/15.2.0/bits/stl_vector.h:1263: constexpr std::vector<_Tp, _Alloc>::ref
erence std::vector<_Tp, _Alloc>::operator[](size_type) [with _Tp = int; _Alloc = std::allocator<int>
; reference = int&; size_type = long long unsigned int]: Assertion '__n < this->size()' failed.

Cause

The error can be traced back to interface.cpp, line 308 in the _update_row_layout function, which has:

sk_interface_set_layout(container_stack.back().layout_widths.size(), &container_stack.back().layout_widths[0], container_stack.back().layout_height);

This line attempts to access the layout_widths vector's first element (container_stack.back().layout_widths[0]), but many callers of _update_row_layout clear layout_widths beforehand, meaning the vector is empty and the out-of-bounds error occurs.

Fix

A simple guard has been added to _update_row_layout, to check if layout_widths is empty before proceeding.

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

Tested primarily using the UI test in sktest.exe in MSYS2 and WSL. Interface displays as expected, with no errors received:

Testing Checklist

• Tested with sktest
• Tested with skunit_tests

Note for Testers

There have been some reported issues running sktest (at all) on MSYS2. Specifically, the animation test was found to loop indefinitely, and some users found the test runner was freezing even before tests were launched. This has been addressed in #132.

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have requested a review from ... on the Pull Request

[222448082Ashen]

General Information

• Type of Change: Bug fix

Code Quality

• Repository: Correct. The PR is made to splashkit-core.
• Readability: High. The addition of the empty check is clear and concise.
• Maintainability: High. This prevents a common source of undefined behavior in C++ (accessing index 0 of an empty vector).

Functionality

• Correctness: The fix is correct. In functions like single_line_layout() and start_custom_layout(), the layout_widths vector is cleared:

  container_stack.back().layout_widths.clear();
  _update_row_layout();

  Without this check, _update_row_layout() would attempt to access &container_stack.back().layout_widths[0], resulting in an out-of-bounds memory access.
• Impact on Existing Functionality: This is a critical stability fix that prevents crashes when using custom or single-line layouts in the SplashKit interface module.

Testing

• Test Results: Logic check confirms that the added condition safely bypasses the layout update when no specific widths are defined, which is the expected behavior for a single-line or custom layout before columns are added.

Documentation

• Documentation: N/A for this minor but important logic fix.

Pull Request Details

• PR Description: Correctly identifies the out-of-bounds risk in the interface module.
• Checklist Completion: All relevant items reviewed.

---

[222448082Ashen]

Recommendations & Observations

1. Robustness: This is a standard defensive programming pattern. By checking .empty() before passing the data pointer of a vector to the backend driver, we ensure the library remains stable even if layout functions are called in an unexpected order.