Test Plan

Test Plan

1. Public API — basic tests to ensure that functions perform as advertized (incl. invalid input)
2. create
3. validate
4. is
5. are_compatible
6. compile 1. flatten 2. unflatten 3. xflatten 4. flatten_msgpack 5. unflatten_msgpack 6. xflatten_msgpack
7. export
8. get_names
9. get_types
10. Frontend
11. Schema definition is loaded — AVRO types, aliases, namespaces, defaults, incl. errors 1. Basic types — null, int, long, float, double, string and bytes 2. array 3. map 4. record 5. fixed 6. enum 7. union 8. namespace handling 9. defaults
12. Data validated correctly against the schema
    • For each AVRO data type check that validation works.
    • Ensure that missing fields are substituted with defaults.
13. Mapping different schema versions work 1. basic types 2. type promotions 3. containers — array, map 4. record-s 5. fixed, matched and mismatched length 6. enums 7. unions — complex rules
14. IL Codegen — generated code yields expected results
15. flatten
16. unflatten
17. xflatten
18. Optimizer — transformations are correct
19. Service Fields
20. known SF types are accepted by compile, generated code behaves as expected
21. unknown SF types rejected by compile
22. some tests for multiple SF-s in [ux]?flatten
23. same but when input and output schemas are different
24. invalid data types in tuple at cells corresponding to SF-s are rejected
25. Hash Collision — STRSWITCH falls back properly when names yield a collision
26. Runtime
27. universal_decode — [ux]?flatten are omnivorous, excercise all possibilities
28. msgpack encoder and decoder routines — can we use msgpuck tests?
29. invalid msgpack input — currently segfaults
30. buffer correctly grown — test with progressively larger arrays up to, say, 10MBs
31. error messages on data validation errors are nice and include location info
32. simulate malloc failures