Add regression tests for matrix inverse

tests/issues/test_issue_4584.odin

@@ -2,6 +2,7 @@
 package test_issues
 
 import "core:testing"
+import "core:log"
 import "core:math/linalg"
 import glm "core:math/linalg/glsl"
 import hlm "core:math/linalg/hlsl"
@@ -91,4 +92,98 @@ test_adjugate_4x4 :: proc(t: ^testing.T) {
 	testing.expect_value(t, hlm.determinant(m), -174)
 	testing.expect_value(t, hlm.adjugate(m) * m, -174 * I)
 	testing.expect_value(t, m * hlm.adjugate(m), -174 * I)
+}
+
+@test
+test_inverse_regression_2x2 :: proc(t: ^testing.T) {
+	I := linalg.identity(matrix[2,2]f32)
+	m := matrix[2,2]f32 {
+		-3, 2,
+		-1, 0,
+	}
+	expected := matrix[2,2]f32 {
+		    0.0,     -1.0,
+		1.0/2.0, -3.0/2.0,
+	}
+	expect_float_matrix_value(t, linalg.inverse(m), expected)
+	expect_float_matrix_value(t, linalg.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * linalg.inverse(m), I)
+
+	expect_float_matrix_value(t, glm.inverse(m), expected)
+	expect_float_matrix_value(t, glm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * glm.inverse(m), I)
+
+	expect_float_matrix_value(t, hlm.inverse(m), expected)
+	expect_float_matrix_value(t, hlm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * hlm.inverse(m), I)
+}
+
+@test
+test_inverse_regression_3x3 :: proc(t: ^testing.T) {
+	I := linalg.identity(matrix[3,3]f32)
+	m := matrix[3,3]f32 {
+		-3,  2, -5,
+		-1,  0, -2,
+		 3, -4,  1,
+	}
+	expected := matrix[3,3]f32 {
+		 4.0/3.0, -3.0,  2.0/3.0,
+		 5.0/6.0, -2.0,  1.0/6.0,
+		-2.0/3.0,  1.0, -1.0/3.0,
+	}
+	expect_float_matrix_value(t, linalg.inverse(m), expected)
+	expect_float_matrix_value(t, linalg.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * linalg.inverse(m), I)
+
+	expect_float_matrix_value(t, glm.inverse(m), expected)
+	expect_float_matrix_value(t, glm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * glm.inverse(m), I)
+
+	expect_float_matrix_value(t, hlm.inverse(m), expected)
+	expect_float_matrix_value(t, hlm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * hlm.inverse(m), I)
+}
+
+@test
+test_inverse_regression_4x4 :: proc(t: ^testing.T) {
+	I := linalg.identity(matrix[4,4]f32)
+	m := matrix[4,4]f32 {
+		-3,  2, -5, 1,
+		-1,  0, -2, 2,
+		 3, -4,  1, 3,
+		 4,  5,  6, 7,
+	}
+	expected := matrix[4,4]f32 {
+		 24.0/29.0, -133.0/87.0,   46.0/87.0,  8.0/87.0,
+		 19.0/58.0,  -46.0/87.0,   5.0/174.0,  8.0/87.0,
+		-35.0/58.0,   71.0/87.0, -55.0/174.0, -1.0/87.0,
+		-11.0/58.0,   16.0/29.0,   -3.0/58.0,  1.0/29.0,
+	}
+	expect_float_matrix_value(t, linalg.inverse(m), expected)
+	expect_float_matrix_value(t, linalg.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * linalg.inverse(m), I)
+
+	expect_float_matrix_value(t, glm.inverse(m), expected)
+	expect_float_matrix_value(t, glm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * glm.inverse(m), I)
+
+	expect_float_matrix_value(t, hlm.inverse(m), expected)
+	expect_float_matrix_value(t, hlm.inverse(m) * m, I)
+	expect_float_matrix_value(t, m * hlm.inverse(m), I)
+}
+
+@(private="file")
+expect_float_matrix_value :: proc(t: ^testing.T, value, expected: $M/matrix[$N, N]f32, loc := #caller_location, value_expr := #caller_expression(value)) -> bool {
+	ok := true
+	outer: for i in 0..<N {
+		for j in 0..<N {
+			diff := abs(value[i, j] - expected[i, j])
+			if diff > 1e-6 {
+				ok = false
+				break outer
+			}
+		}
+	}
+	if !ok do log.errorf("expected %v to be %v, got %v", value_expr, expected, value, location=loc)
+	return ok
 }