GeezerVison

Summary

• Tell us about the problem your pull request is solving.
  • When the query contains DISTINCT or GROUP BY, incorrect total is being returned. This PR fixes it by calculating the count using a subquery. More details below:
  • When DISTINCT clause is used, clearSelect() removes the DISTINCT clause in the countBuilder query, hence returning non-distinct count. For eg:

    • SELECT DISTINCT `users.*` FROM `users` WHERE ....

    • is converted to

    • SELECT `*` FROM `users` WHERE ...

    • losing the distinct property of the query.
  • When GROUP BY clause is used, GROUP BY is applied on COUNT(`${name}.${id}`), which is useless, because rows returned by COUNT will not contain any column other than total to GROUP BY. Hence non-distinct rows are returned again.
  • Hence the solution is, to wrap the main query into a subquery, and count the rows returned by it, which is what this PR does.
• Are there any open issues that are related to this?
  • No. But this is a bug, I can create an issue if necessary.
• Is this PR dependent on PRs in other repos?
  • No