Abstract presentation
Rougail is a robust and powerful, free/open-source configuration manager.
Rougail is:
a CLI (command line interface) utility
a Python library
a YAML based description language
The language combines declaration, data validation, and templating in a single, declarative syntax.
It will be useful to:
define variables (the structure) easily and their constraints
loads those variables
generate variables documentation
loads variables’ values
validate the variables and the overall consistency handling system
generate custom table views
export to different common format (YAML, Ansible, …)
…
Why another configuration manager?
Using Rougail tansforms end user consumer defined consistency rules into highly consistent business objects. It’s a configuration language designed to simplify and unify the way you define, validate and generate configuration data.
In other word, making it easier to manage complex configurations across multiple environments.
You might tell me that other configuration management tools do the same thing. And that’s partly true.
But Rougail adds interesting features in variable management that other projects don’t have. We are of course referring to the documentation management included directly in the configuration manager.
Your configuration is therefore consistent, easily accessible, and modifiable.
Not to mention the always up-to-date documentation and the information you provide to your users regarding changes to variables.
What kind of configuration manager?
At the time of the design of Rougail, there were structuring choices that defined the functioning of the tool.
The steps in Rougail can be explained as follows:
loading the variable’s structured data
loading the user data
reading, validating, exporting, documenting (and so on…) the data
Structured data
Structured data is also called the Rougail format.
- structured data
A variable-first DSL (Domain-Specific Language) designed for describing variables, consistency and describe the relationships between variables in a declarative style.
The language is a mix of YAML and Jinja Templating.
It goes beyond traditional schema languages (like JSON Schema, OpenAPI, Cuelang) by combinig type systems, powerful validation, consistency of the configuration and documentation.
Structured data are commonly placed in structure files.
User data
- user data
User data, as opposed to structured data, are data that only concern the assignment of values and not the consistency of the variables between them.
The variable’s values are also called user values.
The consistency field is outside of the user data scope. The consistency is handled in the structured data‘s scope.
Here a some user data examples:
configuration files
environment variables
external sources
command-line options
a form
Output
- output
Structured and user data form a coherent configuration useful for different purposes.
Output is here to define what kind of data we want.
Here are some output examples:
Tiramisu object
JSON extraction
Ansible inventory extraction
documentation
custom table views
Step |
|---|
Structured data |
User data |
Output |
What kind of actor?
It’s clear that Rougail can be used in many contexts.
Here we’ll define actor names. Obviously, these aren’t the only possible actors.
We’re just defining the actors within the Rougail context. Choice your own actor name if you wish to personnalize in your specific context.
- integrator
An integrator in the Rougail field is the person who writes the structured. He has the responsibility of the integration process, that is, he defines the variables and the relationship between them, the variables that are allowed (or not) to be set, and so on. His responsabilites are the structuration and the consistency of the organisation of the variables between them.
- operator
An operator in the Rougail field is the person who assigns values to the pre-defined variables, his responsabilities are to set variable values correctly.
The user values, that is the values that have been set by the operator, are of course type validated. The type validation is driven by the definitions in the structure file.
Here is a reminder of the different steps:
the integrator defines the structure
the operator sets the value
the integrator and/or the operator use the output
Step |
Actor |
|---|---|
Structured data |
Integrator |
User data |
Operator |
Output |
|
Variable lifecyle
Rougail’s a configuration language and data validation tool is designed to simplify defining, validating, and generating structured configuration and data.
Rougail aims at defining variables.
Here we are talking about the variable lifecyle.
The variable’s lifecyle is the period between its creation and its destruction.
The lifecycle of a variable includes the generic stages (like, in the C language):
Creation: variables are assigned a name and a type
Initialization: they are assigned their first value (we call it default value)
Assignment: the variable’s value is modified
Reading: the variable’s value is used
Destruction: the variable terminates upon the destruction of the object
But other concepts are included in the lifecycle:
Permission: properties describe access constraints
Documentation: informations for variable documentation like description or help. Those informations are used to build documentation, changelog, …
Specialization: define usage, selection,…
Step |
Actor |
Lifecyle |
|---|---|---|
Structured data |
Integrator |
|
User data |
Operator |
|
Output |
|
|
Variable mutability
Structured data
When the integrator defines the structure, variables are mutable.
Even if the default behavior is inconsistent, like when there is a conflict between multiple declarations for the same variable,
it’s possible to explicitly allow to unify (combine) multiple variables declarations.
User data
At the user data level, it’s no more possible to modify any variable definition. Here we are just talking about variable’s values.
Output
Variable definition settings are immutable.
Step |
Actor |
Lifecyle |
Mutability |
|---|---|---|---|
Structured data |
Integrator |
|
Mutable |
User data |
Operator |
|
Immutable |
Output |
|
|
Immutable |
Value access
In the structured data step, there is no value, strictly speaking. The values set are default values. Like other parameters, the default value is explicitly mutable.
In the user data step, we can modify the values. That’s precisely the purpose of this step.
The configuration is said to be in read write mode.
But at the output step, it is obviously no longer possible to modify the value.
The configuration is said to be in read only mode.
Attention
It is important not to confuse value and calculated value.
The result of a calculation can change over time.
In this case, the value does indeed correspond to the result of the calculation.
Step |
Actor |
Lifecyle |
Mutability |
Value |
|---|---|---|---|---|
Structured data |
Integrator |
|
Mutable |
Mutable default value |
User data |
Operator |
|
Immutable |
Read write |
Output |
|
|
Immutable |
Read only |
Access control
Access control is achieved through properties.
- property
A property is a state (
disabled,mandatory,frozen,hidden…) of a family or a variable. These properties change the usual behavior of a variable or family.
The properties can be defined permanently or according to the result of a calculation.
There are two main properties.
Disabled variable
A disabled variable is a variable that will not be accessible to any of the actors (integrator and operator).
This property is generally used dynamically to remove access to the variable depending on the context.
A picture is worth a thousand words:
Step |
Actor |
Lifecyle |
Mutability |
Value |
Access control |
|---|---|---|---|---|---|
Structured data |
Integrator |
|
Mutable |
Mutable default value |
N/A |
User data |
Operator |
|
Immutable |
Read write |
|
Output |
|
|
Immutable |
Read only |
|