1.2 Document Conventions
1.2 Document Conventions
Requirement Levels (RFC 2119)
This specification uses the keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD, SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL as defined in RFC 2119 to indicate requirement levels:
| Keyword | Meaning |
|---|---|
| MUST / REQUIRED / SHALL | Absolute requirement |
| MUST NOT / SHALL NOT | Absolute prohibition |
| SHOULD / RECOMMENDED | Strong recommendation; deviation requires a justified reason |
| SHOULD NOT / NOT RECOMMENDED | Strong recommendation against; use only with a justified reason |
| MAY / OPTIONAL | Truly optional |
The keywords are normative only when they appear in uppercase.
Notation
| Convention | Form | Example |
|---|---|---|
| JSON example | Fenced code block with json highlighting | "lexId": "1" |
| Placeholder | Angle brackets | <lexId> |
| Optional element | Marked [optional] | field [optional] |
| Repetition | Three dots | ..., ... |
| Pattern | Regular expression (ECMA-262) | ^[1-9][0-9]*$ |
| Field type | Refers to types in Chapter 3.2 | UUID, DateTime |
Naming Conventions
RUML distinguishes schema-defined fields from user-defined identifiers by case style:
- Schema-defined fields use camel case — e.g.
lexId,id,children,containerPath,validFrom,documentType,decisionMode,expression,onTrue. - User-defined identifiers use snake case — e.g. DataPoint IDs (
dp_annual_income), Context IDs (ctx_court_ruling), and Text IDs (text_bgb_823) (Rule DP-02DP-02 The key of a DataPoint, the id of a set and the keys of a set'sfieldsMUST match the pattern[a-z_][a-z0-9_]*.). - Implementation-defined identifiers, such as references to actions or AI models, SHOULD use
snake case — e.g.
document_service,manual_review_workflow.
Tools SHOULD follow this convention when generating new identifiers. Capability IDs follow their own namespace convention defined in Chapter 4.4.