Running functions
When an operation requires more than just the schema, and data is necessary, the dynamic logic can be built into a separate tool. Functions bundle dynamic logic in container images and apply that logic to the contents of a package – modifying and validating package contents.
Functions provide a common interface for writing programs to read and write resources as data. This enables greater reuse and composition than when configuration is itself represented as code. Functions may be written in any language, or simply wrap other existing programs.
Let’s look at the example of imperatively running a function to set a label value. The label-namespace image contains a program which will modify the
kpt fn run --image gcr.io/kpt-functions/label-namespace . -- label_name=color label_value=orange
Kpt read the resources from the package at “.”, provided them as input to
a container running gcr.io/kpt-functions/label-namespace, and wrote the
resources back to the package.
This particular function (label-namespace) adds a label to all Namespace resources provided to it.
Functions can address many types of workflows, including:
- Generate resources from some inputs (like client-side CRDs)
- Apply cross-cutting transformations (e.g., set a field on all resources that look like this)
- Validating resources conform to best practices defined by the organization (e.g., must specify tag as part of the image)
See the functions catalog for more examples of functions, and the [function docs] for more details on writing functions.