All of the examples of using

BigQueryResource

show the bigquery client being instantiated as a context manager from

bigquery.get_client()

. This differs the BigQuery API docs, where the

bigquery.Client()

can be instantiated and used without the context manager. Why is this? Is a context manager required for the

BigQueryResource

?

The use of a context manager with

BigQueryResource

in the examples provided in the documentation is a design choice that aligns with best practices for managing resources in Python. When you use a context manager, it ensures that resources are properly allocated and released, which is particularly important for network connections and clients that interact with external services like Google BigQuery. The context manager pattern, often used with the

with

statement in Python, is designed to handle setup and teardown operations automatically. In the case of

BigQueryResource

, the context manager is used to acquire a client and ensure that it is properly closed after use, even if an error occurs during the operation. This helps prevent potential issues such as leaking connections or not properly cleaning up resources. While the BigQuery API allows for the client to be instantiated and used without a context manager, the use of a context manager in the

BigQueryResource

examples is a way to encourage good resource management practices within the Dagster framework. It is not strictly required to use a context manager, but it is recommended to handle resources safely and efficiently. The documentation provides an example of using

BigQueryResource

with a context manager as follows:

python @asset def small_petals(bigquery: BigQueryResource):     with bigquery.get_client() as client:         return client.query(             'SELECT * FROM IRIS.IRIS_DATA WHERE "petal_length_cm" < 1 AND'             ' "petal_width_cm" < 1',         ).result()

In this example, the

get_client

method of

BigQueryResource

is used within a

with

statement, which takes care of opening and closing the client properly. This pattern is a common Pythonic way to manage resources that require clean-up after their use.