Question

I have two tables:

table1:

id(int) | stuff(text)
-------------------------
1       | foobarfoobarfoo
2       | blahfooblah
3       | foo

table2:

id(int) | otherstuff(text)
--------------------------
1       | foo
2       | bar
3       | blah

A row in table1 can have more than one of foo, bar etc. And, each row in table2 can appear in more than one row of table1.

Which is a better way of keeping this straight. Should I create a third table like this:

table3:

id_from2(int) | id_from1(int)
-----------------------------
1             | 1
1             | 2
1             | 3
2             | 1
3             | 2

Or, should I have an column of type array added to table1 and table2 to keep track of the same information?

Answer 1

Yes, using junction tables is the correct way of implementing many-to-many relations in RDBMS.

You can add more attributes to your junction table (i.e. table3) if necessary. For example, if the relations are ordered, you can add a third field that specifies an ordering of the (table1, table2) combinations. Here is a link to an answer on Stack Overflow that gives a nice detailed example of a many-to-many table.