Policies and configuration
Transformation controls are independent unless stated otherwise. In particular,
correct=True or --correct does not enable or replace the explicit policies.
Legacy correction behavior
The correct option defaults to False. When enabled, it applies the existing
class and property semantic checks and corrections, including:
reconciling class stereotypes with
isExtensional,isPowertype, andordervalues;assigning corrected default class order after an incompatible value is removed; and
checking stereotyped properties against their owning class and assigning the
eventstereotype when that established rule applies.
Missing vocabulary-defined default attributes are supplied whether or not
correct is enabled. Explicit policies for invalid stereotypes, cardinalities,
and unresolved references also run independently of it.
Invalid stereotypes
Assigned Class, Relation, and Property stereotypes are normalized to lowerCamelCase before validation. A lexical variant that normalizes to a valid stereotype is emitted canonically and produces a normalization warning.
Value |
Consequence |
|---|---|
|
Warn and emit the normalized stereotype triple even when it is invalid for the element type. |
|
Warn and omit only the invalid stereotype triple. |
|
Raise an error and abort before file output. |
Invalid cardinalities
Valid source forms are *, a non-negative integer, n..m, and n..*, with a
lower bound no greater than a finite upper bound. * is normalized to 0..*,
and an integer n is normalized to n..n.
Value |
Consequence |
|---|---|
|
Preserve the original |
|
Repair only known separator forms ( |
|
Raise an error and abort before file output. |
Successful repair emits the repaired value and bounds with a repair warning.
Valid and successfully repaired lower and upper bounds use
xsd:nonNegativeInteger.
Unresolved diagrammatic model references
This policy applies only to modelElement references on objects contained by
diagrams. It does not validate other reference categories such as source,
target, propertyType, general, or specific.
Value |
Consequence |
|---|---|
|
Warn, omit the unresolved |
|
Warn, preserve the relation, and materialize the referenced target. |
|
Raise an error and abort before file output. |
Path-point order
The policy is evaluated only for complete-project output and paths with more than one point.
Value |
Consequence |
|---|---|
|
Emit point triples without order and warn that the sequence is absent. |
|
Also add the coordinate sequence as a non-normative |
The comment is explanatory text, not a vocabulary-defined ordering structure.
Property assignments
Only non-empty propertyAssignments maps on resources present in the resulting
graph are affected. Null and empty maps are ignored.
Value |
Consequence |
|---|---|
|
Omit the map and warn. |
|
Add canonical JSON as a non-normative |
Canonical JSON sorts object keys and preserves array order.
Resource identity
Configuration |
Effective namespace |
|---|---|
No base URI (default) |
|
Exact base URI |
The supplied absolute URI, normalized with an identifier separator when needed |
Parent base URI plus content ID |
|
The CLI exposes the latter two modes as mutually exclusive options. The Python
API uses base_uri and append_content_hash.
Transformation provenance
Value |
Consequence |
|---|---|
|
Return or write only the model graph. |
|
Add provenance to the same graph; its generation timestamp makes the result non-deterministic. |
|
CLI only: keep the model file unchanged and write |
The provenance records the source filename and SHA-256 identifier, software
version, requested and effective identity configuration, output format, and
output-affecting transformation options. It does not record source or output
paths, silent, or batch orchestration.