# bf-3m477: Test Each Timing Flag and Capture Output

## Task
Test each timing flag and capture output for analysis.

## Implementation

### Flags Tested

All 5 timing flags from bf-4wz6v documentation were tested:

1. **`--message-format=libtest-json-plus`** ✅
   - Command: `NEXTEST_EXPERIMENTAL_LIBTEST_JSON=1 cargo nextest run -p pdftract-core --lib --message-format=libtest-json-plus`
   - Output: `notes/bf-mge0o-message-format-libtest-json-plus.log` (2.0M)
   - Result: SUCCESS - JSON output with `exec_time` fields per test

2. **`--message-format-version=0.1`** ✅
   - Command: `NEXTEST_EXPERIMENTAL_LIBTEST_JSON=1 cargo nextest run -p pdftract-core --lib --message-format=libtest-json-plus --message-format-version=0.1`
   - Output: `notes/bf-mge0o-message-format-version-0.1.log` (2.0M)
   - Result: SUCCESS - JSON output with versioned schema (version must be in `major.minor` format, major must be 0)

3. **`--status-level=slow`** ✅
   - Command: `cargo nextest run -p pdftract-core --lib --status-level=slow`
   - Output: `notes/bf-mge0o-status-level-slow.log` (112K)
   - Result: SUCCESS - Standard output format, marks slow tests during execution

4. **`--final-status-level=slow`** ✅
   - Command: `cargo nextest run -p pdftract-core --lib --final-status-level=slow`
   - Output: `notes/bf-mge0o-final-status-level-slow.log` (412K)
   - Result: SUCCESS - Standard output format with slow test summary at end

5. **`--message-format=libtest-json`** ✅
   - Command: `NEXTEST_EXPERIMENTAL_LIBTEST_JSON=1 cargo nextest run -p pdftract-core --lib --message-format=libtest-json`
   - Output: `notes/bf-mge0o-message-format-libtest-json.log` (2.0M)
   - Result: SUCCESS - JSON output without nextest-specific extensions

### Key Findings

1. **Experimental flag requirement**: Both `libtest-json` and `libtest-json-plus` require `NEXTEST_EXPERIMENTAL_LIBTEST_JSON=1` environment variable
2. **Version format**: `--message-format-version` requires `major.minor` format (e.g., `0.1`), not just `1`
3. **Version constraints**: Major version must be 0 (range 0..1)
4. **JSON format**: Both JSON formats produce per-test `exec_time` fields for timing data
5. **Slow test flags**: `--status-level=slow` and `--final-status-level=slow` work without experimental flags

### Test Suite
- 2,902 lib tests from `pdftract-core`
- Runtime: ~4.8s per run
- 2857 passed, 45 failed, 2 skipped

## Acceptance Criteria

- ✅ **PASS**: Output file exists for each tested flag (5 files created)
- ✅ **PASS**: Each file is named after the flag it tests
- ✅ **PASS**: At least 3 flags are tested (5 tested, exceeds requirement)

## Output Files Created

1. `notes/bf-mge0o-message-format-libtest-json-plus.log` (2.0M)
2. `notes/bf-mge0o-message-format-version-0.1.log` (2.0M)
3. `notes/bf-mge0o-status-level-slow.log` (112K)
4. `notes/bf-mge0o-final-status-level-slow.log` (412K)
5. `notes/bf-mge0o-message-format-libtest-json.log` (2.0M)

## References

- Third child of split bf-mge0o
- Depends on bf-2n36a (baseline comparison)
- Documentation from bf-4wz6v timing flag research

## Verification

All timing flags successfully executed and output captured. The JSON formats (libtest-json and libtest-json-plus) both provide per-test timing data via the `exec_time` field, making them suitable for programmatic consumption of test execution times.