Sharing Data

The value of include is in its simplest form a string which represents a path to a JSON or YAML file in the project. Multiple files can be referenced by an array of strings. The include keyword is necessarily embedded in a map. The top level element of the referenced file must be a map. This is necessary so the deep-merge operation discussed on the composing data can be applied.

We use the symbol ⊕ to denote deep-merge in the following. Let the include statement be contained in the map M. The maps of the included files are denoted with M1, M2, and so forth where the number corresponds to the index in the array. Then the resulting map is given by M1 ⊕ M2 ⊕ ... ⊕ M. The key observation is that conflicting keys with higher index override those with lower index and that keys of the map with the include statement have the highest precedence of all.

The include directive may appear in any map at any level. Further, files being included may contain include statements, too. The merging strategy will by applied recursively in DFS order until all include statements have been replaced. It is illegal to define circular references, they will result in an error.

Configuration data can be shared between projects via git submodules. To specify the submodule path the include statement must be given in its full form: an array of maps each containing a path and submodule key. The value of the path is string representing the file in the project or submodule. The value of the submodule is an array of strings. Each string represents the path to the submodule in the project. The array enables references to nested submodules.

Every referenced submodule must be configured as a project in the Cider-CI instance and the referenced commit must be known by the Cider-CI instance.