Numeric and Unicode Edge Vectors
Standard: SCS-TEST-002 | Edition 1.0 | Status: executable cross-language batch
Purpose
This suite extends the arithmetic/ASCII smoke test with explicitly defined fixed-width integer and Unicode observations. It currently tests C11, C++17, and Java 21+ on the configured operating systems. It does not yet claim SLeeLa VM coverage for these edge cases.
Normative expectations
- Signed 32-bit minimum and maximum are exactly -2147483648 and 2147483647.
- Unsigned 32-bit addition wraps modulo 2^32; UINT32_MAX + 1 is zero.
- Signed integer division truncates toward zero for the selected operands: -7 / 3 is -2.
- Signed remainder follows the quotient identity for the selected operands: -7 % 3 is -1.
- The string Aé😀 encodes as seven UTF-8 bytes and contains three Unicode code points.
- A byte count, UTF-16 code-unit count, Unicode code-point count, and grapheme count are different measurements. The suite explicitly reports UTF-8 byte length and Unicode code-point count, not a generic “string length”.
Vector ledger
| ID | Input / operation | Expected |
|---|---|---|
| INT32-MIN | Minimum signed 32-bit value | -2147483648 |
| INT32-MAX | Maximum signed 32-bit value | 2147483647 |
| UINT32-WRAP | UINT32_MAX + 1 | 0 |
| SIGNED-DIV | -7 / 3, truncating toward zero | -2 |
| SIGNED-REM | -7 % 3 | -1 |
| UTF8-BYTES | UTF-8 byte length of Aé😀 | 7 |
| UNICODE-CODEPOINTS | Unicode code-point count of Aé😀 | 3 |
Implementation notes
- C uses int32_t/uint32_t and a small UTF-8 leading-byte/continuation-byte counter for this known-valid test string.
- C++ uses std::int32_t/std::uint32_t and a UTF-8 byte string; it counts code points by UTF-8 sequence width for this known-valid test string.
- Java uses int, Integer.MIN_VALUE/MAX_VALUE, StandardCharsets.UTF_8, and String.codePointCount.
- The C/C++ UTF-8 helpers are intentionally scoped to the known-valid fixture. They are not general-purpose Unicode validators and do not test malformed, overlong, surrogate, or out-of-range sequences.
Commands
Run locally with Python 3, a C11 compiler, a C++17 compiler, and JDK 21+:
python3 comparison/tests/run_edge_vectors.py
The runner emits comparison-edge-results.json with actual outputs, toolchain version strings, platform, and overall status. CI uploads this file per operating system.
Qualification boundary
Passing this suite qualifies only these seven observations for the tested compiler/runtime versions and CI platforms. It does not establish SLeeLa VM behavior, floating-point equivalence, every integer width, overflow rules for signed values, Unicode normalization, grapheme segmentation, malformed UTF-8 handling, or ABI compatibility.