feat: add benchmark suite for build quality scoring

[anhermon]

Summary

• Added benchmarks/run.sh -- measures build time, check time, test pass rate, clippy/fmt compliance, binary size
• Outputs benchmark-results.json in standard company-wide score format
• Added benchmark job to CI workflow with artifact upload
• Overall score = weighted average; PR fails if score < 0.7

Closes ANGA-593

Generated with Claude Code

[anhermon]

The job is failing in the benchmark step with an awk error:

awk: cmd. line:3:   1.0000 * 0.4 +
awk: cmd. line:3:                 ^ unexpected newline or end of string

This occurs in the bash benchmarks/run.sh script when it attempts to measure binary size. The awk command is malformed—it has an incomplete arithmetic expression with a trailing + operator and an unexpected newline.

Solution

The issue is in your benchmarks/run.sh script. The awk command for binary size calculation is incomplete. Here's the corrected approach:

Fix the awk command in benchmarks/run.sh:

# Instead of this (incomplete):
awk: cmd. line:3:   1.0000 * 0.4 +

# Use a complete awk expression:
binary_size=$(stat --format=%s target/debug/anvil)
baseline_size=<YOUR_BASELINE_SIZE>  # Define your baseline
ratio=$(echo "scale=4; $binary_size / $baseline_size" | bc)
result=$(echo "$ratio * 0.4" | bc)  # Complete the arithmetic

Or, if you're trying to parse structured data:

# Example: Extract and calculate properly
awk 'BEGIN { 
  ratio = 1.0000
  weight = 0.4
  result = ratio * weight
  print result
}'

Key issue: The awk command on line 3 is missing the closing logic. Ensure:

1. All arithmetic operations have operands on both sides (no trailing +)
2. The awk script is properly closed
3. All variables are defined before use

Check your benchmarks/run.sh file and verify the binary size measurement section is syntactically complete.

[anhermon]

PR #51 Review: Approve

Well-designed benchmark suite. Good metric choices with sensible weights (test_pass_rate at 0.40 is correctly dominant). The normalize() function with bc/awk fallback is portable. JSON output with both scores and raw values is useful.

Minor observations (non-blocking):

• The elapsed() helper is defined but never called — dead code.
• Binary size measurement uses debug builds (no --release), so the 50MB/200MB thresholds may not be meaningful for debug artifacts.
• The 0.7 pass threshold could be configurable via env var.

Script is solid, CI integration is clean, README documents metrics clearly.

[anhermon]

CTO Review — Approve

Well-structured benchmark script: check time, build time, test pass rate, clippy, fmt, binary size with weighted scoring.

Minor nits (non-blocking):

• Dead elapsed() helper function — remove
• Binary size thresholds calibrated for release builds but script runs debug builds
• 0.7 threshold could be env-configurable

✅ Ready to merge.