Pipeline developer reference

This section contains reference and instructions for pipeline developers.

Data loading and transformations are implemented in pipelines by queries that define streaming tables and materialized views. To implement these queries, Lakeflow Spark Declarative Pipelines supports SQL and Python interfaces. Because these interfaces provide equivalent functionality for most data processing use cases, pipeline developers can choose the interface that they are most comfortable with.

Python development

Create pipelines using Python code.

Topic	Description
Develop pipeline code with Python	An overview of developing pipelines in Python.
Lakeflow Spark Declarative Pipelines Python language reference	Python reference documentation for the pipelines module.
Manage Python dependencies for pipelines	Instructions for managing Python libraries in pipelines.
Import Python modules from Git folders or workspace files	Instructions for using Python modules that you have stored in Databricks.

SQL development

Create pipelines using SQL code.

Topic	Description
Develop pipeline code with SQL	An overview of developing pipelines in SQL.
Pipeline SQL language reference	Reference documentation for SQL syntax for Lakeflow Spark Declarative Pipelines.
Use pipelines in Databricks SQL	Use Databricks SQL to work with pipelines.

Other development topics

The following topics describe other ways to develop piplines.

Topic	Description
Convert a pipeline into a Databricks Asset Bundle project	Convert an existing pipeline to a bundle, which allows you to manage your data processing configuration in a source-controlled YAML file for easier maintenance and automated deployments to target environments.
Create pipelines with dlt-meta	Use the open source dlt-meta library to automate the creation of pipelines with a metadata-driven framework.
Develop pipeline code in your local development environment	An overview of options for developing pipelines locally.