We define the objects in our system with a 'State Based Approach'.
We then find a way to show user balances for a group: summing group expenses per user.
When finding the overall balances in a group, why didn't we use a "group by" clause to sum all the user balances from the expense table?
The number of users in a group is variable. We could try to denormalize the expenses table into two tables of balances and expense_info:
balances:
expense_id
user_id
balance
paid
owes
expense_info:
id
title
desc
imageUrl
group_id
On a 'getGroupExpenses' request for group id=123, we fire the database query:
'select user_id, sum(balance) from balances where expense_id in (select id from expense where group_id = '123') group by user_id'
This requires a nested query, which could perform poorly.
However, this style of querying will perform very well for finding an individual user's expenses.
Depending on what you are trying to optimize, you may not may not denormalize the tables.