splunk/qbec v0.13.4

on GitHub

Note: some of the features in this release have been removed in v0.14.0. Please refer to those release notes as well.

a.k.a the "scale" release. This release mainly contains enhancements that allow qbec to be used at scale on
large monorepos and/ or when deploying to tens of clusters.

This release has no backwards-incompatible changes.

• update the glob importer to support ** patterns. Glob references are now matched using the
  doublestar library
• Add a native function labelsMatchSelector to expose K8s label matching in jsonnet code
• Allow componentsDir to be a glob as opposed to a single directory. This change makes qbec load components
  from potentially multiple directories but does not introduce any namespace semantics. Components must still be
  uniquely named across all such directories. It is only useful for compartmentalizing files in a monorepo
  where different directories have different code owners.
• Add support for pre-processors. A pre-processor is some jsonnet code that is evaluated before any components is.
  The return value of every such preprocessor is set as the external code variable called computed.qbec.io/<base-name>
  where <base-name> is the base name of the preprocessor file without its extension. This variable is available for
  use in component code. This allows for caching values that are costly to compute.
• qbec now sets the qbec.io/component external variable when it evaluates components, pre- and post-processors.
  This may be used in component code to key into a params object, for example. For pre- and post-processors,
  the component name is set to <pre|post>processor.qbec.io/<base-name> where <base-name> has the same semantics
  as described above.
• Allow multiple pre- and post-processors. The qbec.yaml attributes, preProcessor and postProcessor although
  singular, can be set to a list of files separated by :. Attribute names will be fixed to be in plural form and
  accept an array of strings in a future, backwards-incompatible release.