allocate memory only once

The best_patterns is only an array of arrays and hence does not need
that much memory. Overestimate the size by assuming we have no duplicate
columns and allocate memory only once. This makes the code more readable
and probably faster too (since reallocating takes time too).

src/cutting_stock.c

@@ -135,8 +135,6 @@ double **cutting_stock_compute_best_patterns(order **orders, int order_count, in
     columns_matrix = columns_matrix_compute(orders, order_count, max_width);
     column_size = order_count;
     columns_matrix_number = order_count;
-    best_patterns = NULL;
-    temp_best_patterns_number = 0;
 
     while (true) {
         dual_column = cutting_stock_compute_dual(orders, order_count, columns_matrix, columns_matrix_number);
@@ -164,18 +162,14 @@ double **cutting_stock_compute_best_patterns(order **orders, int order_count, in
         return NULL;
     }
 
+    best_patterns = (double **)malloc(columns_matrix_number * sizeof(double *));
+    temp_best_patterns_number = 0;
     for (i = 0; i < columns_matrix_number; i++) {
-        if (best_patterns) {
-            if (!double_matrix_contains_array(best_patterns, temp_best_patterns_number, columns_matrix[i], column_size)) {
-                best_patterns = (double **)realloc(best_patterns, (temp_best_patterns_number + 1) * sizeof(double *));
-                best_patterns[temp_best_patterns_number] = columns_matrix[i];
-                temp_best_patterns_number++;
-            }
-        } else {
-            best_patterns = (double **)malloc(sizeof(double *));
-            best_patterns[0] = columns_matrix[i];
-            temp_best_patterns_number++;
-        }
+        if (double_matrix_contains_array(best_patterns, temp_best_patterns_number, columns_matrix[i], column_size))
+            continue;
+
+        best_patterns[temp_best_patterns_number] = columns_matrix[i];
+        temp_best_patterns_number++;
     }
 
     *best_patterns_number = temp_best_patterns_number;