STACK compliance
Teaching implications first, followed by comparison details
STACK authors and maintainers reviewing conventions can go directly to semantics questions and evidence.
For teachers
Each case asks whether a particular STACK answer test accepts a submitted answer. This report shows which cases STACK SymPy can check, where its result agrees with STACK, and where a teacher should expect a difference.
How to read the results
Port means STACK SymPy. STACK means the reference Moodle/Maxima answer test. Expected is the outcome recorded in the original case. Green can mean that a wrong answer was correctly rejected. A matching result for one case does not establish that every question using that test behaves identically.
State the mathematical domain, distinguish written form from value, and use numerical tests for approximate answers. The semantics discussions for STACK authors record the detailed conventions and unresolved questions. Start with the authoring guide when preparing a question, and run its own tests as well as inspecting the examples. Python answer tests for matrix tasks are documented separately from this native comparison.
Green: expected behavior or agreement Amber: known issue or explained difference Red: unexpected result Gray: not graded, outside scope, or unavailable
More comparisons: final marks, test applicability, and STACK expectations
Cases
Select a count above or use the filters. Open a case to see the expected and actual results.
For developers
Scope and interpretation
The inventory contains 2,303 imported answer-test cases. It compares CAS functions, not complete Moodle grading sessions. The port currently provides ten native-compatible answer-test families. Unsupported cases remain visible by missing test or input feature; some direct STACK calls require parser or controller preparation and are not compared.
“Test applicability” asks whether a test can compare the supplied values. It is separate from syntax validity and the accept/reject mark. Form tests may reject equivalent expressions. Browser agreement includes invalid/unsupported classifications as well as graded results; the summary distinguishes them.
Source codes 0 and 1 request rejection and acceptance; -1 requests a controlled error. Codes -2 and -3 flag source-authored native deviations. Only reviewed cases receive the “Known STACK issue” label. Ordinary notes are retained but not generally compared; asserted controlled-error notes are checked separately.
The Python-only PythonChecker test, including its supplied basis and matrix property functions, does not increase the native inventory or its coverage counts. Its authored, unit and browser checks establish separate Python behavior. See the developer reference.
Saved evidence and reproduction
The detailed conventions reference records case-specific reasoning, sources and open questions. The development guide explains how to reproduce the reports. A normal website render reads the saved results; it does not rerun STACK comparisons.