Currently, you build and run the tests and if they fail, manually delete the generated sql, add additional instructions in function spec and then run it again. A self heal loop can support two types of auto-fixes
- If the SQL generated has syntax errors - it can be fixed within the loop itself
- If the SQL has functional error and generated incorrect data causing the test to fail, the newly generated sql shouldn't be cached - so a rerun of the test regens the SQL.
Currently, you build and run the tests and if they fail, manually delete the generated sql, add additional instructions in function spec and then run it again. A self heal loop can support two types of auto-fixes