Loops

The loop function in DUQL is used for iterative processing. It applies a sequence of steps repeatedly to an initial dataset until a termination condition is met, typically when the step function returns an empty table.

Syntax

loop:
- <step_1>
- <step_2>
# ...

Parameters

The loop function doesn't have parameters itself, but it contains a list of steps that will be executed iteratively until the input (the results from the previous step) is empty.

Component	Type	Required	Description
steps	array	Yes	A list of transformation steps to be applied iteratively

Each step in the loop can be any valid DUQL transformation, such as filter, generate, group, etc.

Examples

Basic Iterative Calculation

loop:
- filter: remaining_balance > 0
- generate:
    remaining_balance: remaining_balance - payment_amount
- append: payments_made

This loop continues to process payments until the remaining balance is zero or negative.

Hierarchical Data Processing

loop:
- join:
    dataset: employees
    where: managers.id == employees.manager_id
- generate:
    level: level + 1
- filter: not is_null(manager_id)

This loop traverses an employee hierarchy, joining the manager table with itself until it reaches employees with no manager (top level).

Use Cases

1. Recursive Calculations: Perform calculations that depend on previous results, such as compound interest or depreciation.

2. Hierarchical Data Processing: Traverse tree-like structures, such as organizational hierarchies or bill of materials.

3. Iterative Data Cleaning: Apply data cleaning steps repeatedly until certain quality criteria are met.

4. Convergence Algorithms: Implement algorithms that iterate until a convergence condition is satisfied.

Best Practices

1. ⚠️ Always include a termination condition to prevent infinite loops. This is typically done using a filter step that will eventually return an empty result.

2. 🔢 Consider adding a maximum iteration count as a safeguard against unexpected infinite loops.

3. 📊 Use generate steps within the loop to create or update variables that track the iteration progress or accumulate results.

4. 🧮 When possible, try to express your logic without loops for better performance. Only use loops when iterative processing is truly necessary.

5. 📝 Document the purpose and expected behavior of your loop clearly, especially for complex iterative processes.

Related Functions

• filter: Often used as a termination condition in loops
• generate: Used to update variables within the loop
• append: Useful for accumulating results across iterations

Limitations and Considerations

• Loops can be computationally expensive, especially on large datasets. Use them judiciously.

• Not all database systems support iterative processing natively.

• Complex loops can be difficult to optimize. Consider alternative non-looping approaches if performance becomes an issue.

---

💡 Tip: While loops are powerful for certain types of problems, they should be used sparingly in data processing pipelines. Often, set-based operations are more efficient and easier to optimize. Always consider if there's a non-iterative way to express your logic before resorting to a loop!