I'm currently exercising with database normalization, and I find that a lot of sources differ in approach to get to 1NF.
For example, this is my UNF table:
customer
+----+--------+----------------------+
| id | name | phone |
+----+--------+----------------------+
| 1 | achmed | 06-101010, 06-111111 |
+----+--------+----------------------+
| 2 | jozef | 06-232323 |
+----+--------+----------------------+
| 3 | maria | 06-464646, 06-989898 |
+----+--------+----------------------+
One approach splits the multi-values into different tuples, which creates redundancy temporary:
customer
+----+--------+-----------+
| id | name | phone |
+----+--------+-----------+
| 1 | achmed | 06-101010 |
+----+--------+-----------+
| 1 | achmed | 06-111111 |
+----+--------+-----------+
| 2 | jozef | 06-232323 |
+----+--------+-----------+
| 3 | maria | 06-464646 |
+----+--------+-----------+
| 3 | maria | 06-989898 |
+----+--------+-----------+
Another approach splits the multi-values directly into new relations, which could look like this:
customer
+----+--------+
| id | name |
+----+--------+
| 1 | achmed |
+----+--------+
| 2 | jozef |
+----+--------+
| 3 | maria |
+----+--------+
customer_phone
+----+-----------+
| id | phone |
+----+-----------+
| 1 | 06-101010 |
+----+-----------+
| 1 | 06-111111 |
+----+-----------+
| 2 | 06-232323 |
+----+-----------+
| 3 | 06-464646 |
+----+-----------+
| 3 | 06-989898 |
+----+-----------+
Both will end up similar after higher normal forms, but which approach should be 'the best practice' as 1NF and why?