“Webster collects definitions and writes them down.” – Noah Webster Jr., 1928, whilst hulking out on the first edition of his An American Dictionary of the English Language
Most features of OakPAL are focused on testing FileVault content packages as part of build and deployment pipelines, but OakPAL Webster is designed as a tool to help manage and curate the definitions that compose the initial JCR state that content packages and OakPAL checklists depend on.
Whereas OakPAL reacts to a transient built artifact, building a pristine Oak repository for repeatable execution with quick iteration cycles, Webster is intended for infrequent manual execution, and it reads and writes files that are normally committed to git with the rest of your application source code, while incorporating JCR state from the content repositories of locally-installed Oak-based applications like Sling or AEM.
This initial JCR state encompasses these definitions:
Namespaces: In JCR, namespaces are represented by both a prefix and a URI. Namespaces are a common dependency of any fully-qualified JCR definition, like a node type, privilege, or qualified node name. In some cases, an existing uri can be remapped to a different prefix, but this is rare and not always allowed. OakPAL attempts to support remapping wherever possible, but best practice recommends that you treat the prefix as a globally standardized attribute of the URI.
Privileges: JCR defines some basic privileges for use in Access Control, like jcr:read and rep:write. Some applications, like AEM, define additional privileges, like crx:replicate. OakPAL doesn’t currently evaluate Access Control restrictions, but it does require that any privileges referenced by rep:policy nodes are registered by a checklist or by a content package containing a META-INF/vault/privileges.xml file.
Node Types: Every node in a JCR repository must have a primary type (a subtype of nt:base) and zero-or-more mixin types. The node type is a critical definition to reproduce accurately during an OakPAL execution, because it restricts where a node can be installed in the repository, and it can restrict the property names and values set on the node when a package import attempts to save the change. Content packages can provide required node type definitions in a META-INF/vault/nodetypes.cnd file that will be used only to register new definitions, without redefining existing definitions.
Root Nodes: Also known as “Structure Nodes”, root nodes, when appropriately typed, may contribute to initial JCR state required for particular checks. OakPAL Checklists can arbitrarily declare that a node of any defined type can exist at any arbitrary path. This allows for testing a content package that might attempt to add children to that node, in order to enforce node type constraints during a dry run.
OakPAL assumes that built-in Oak definitions are present in all execution contexts, and therefore not exported to sidecar definition formats like nodetypes.cnd, privileges.xml, and checklist.json.