JSON arrays that originate from unordered Python sets are serialized in non-deterministic order. Affected are the top-level object lists and set-valued attributes such as submodel and isCaseOf.
Expected behavior:
Serializing the same AAS twice produces byte-identical JSON, so output is reproducible across runs.
Actual behavior:
Array element order varies between runs. Set iteration order depends on hash randomization (PYTHONHASHSEED), so each serialization can emit a different ordering. This makes output non-reproducible — complicates diffing, caching, and snapshot testing.
JSON arrays that originate from unordered Python sets are serialized in non-deterministic order. Affected are the top-level object lists and set-valued attributes such as submodel and isCaseOf.
Expected behavior:
Serializing the same AAS twice produces byte-identical JSON, so output is reproducible across runs.
Actual behavior:
Array element order varies between runs. Set iteration order depends on hash randomization (PYTHONHASHSEED), so each serialization can emit a different ordering. This makes output non-reproducible — complicates diffing, caching, and snapshot testing.