Skip to content

Imprecise tests: Python data types vs. YANG data types #370

Description

@dthiele

Some checks and tests are performed based on Python data types. Python and YANG, however, have different equality semantics for certain data types. One such example are integers and booleans. In Python, the following comparison yields True.

>>> {"boolean": 1} == {"boolean": True}
True

While in a YANG, according to https://datatracker.ietf.org/doc/html/rfc7951#section-6.3, this would not be the case.

As many tests are based on Python's equality operator (e.g. assertEqual()), there is a potential gag in test coverage. Here is a concrete example:

Run original json-serialize tests

tox -- tests/serialise/json-serialise/

Modify the golden sample to violate https://datatracker.ietf.org/doc/html/rfc7951#section-6.3 (e.g. change the "boolean" leaf node from "true" to "1"), print the entire file, and rerun the test.

# The following commands require the jq and sponge tools
jq '.c1.l1."1".boolean=1' tests/serialise/json-serialise/json/expected-output.json | sponge tests/serialise/json-serialise/json/expected-output.json
jq . tests/serialise/json-serialise/json/expected-output.json
tox -- tests/serialise/json-serialise/

Despite the modification, tests/serialise/json-serialise/ returns no issues.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions