Harvest Timing Fixture: Node, Python, and azd
By FrootAI Engine Team · CC0-1.0 · Source declared: Pre-filled sample-data.json timings; no runner logs or Azure OpenAI execution receipt is attached
Not citation-grade evidence
The script replays pre-filled timings. No 50-repository corpus, run logs, environment receipt, raw timing samples, or implementation-parity evidence is attached.
Do not use this fixture to claim cost, carbon, performance, or product superiority.
Fixture implementations
3
Results by Category
Milliseconds (median, Node implementation, n=50 repos)
Milliseconds (median, Node implementation, n=50 repos)
View data table
| Category | Node pipeline |
|---|---|
| S1 Discover | 320 |
| S2 Fetch | 480 |
| S3 Extract | 1,200 |
| S4 Retrieve | 650 |
| S5 Scaffold | 980 |
| S6 Compose | 380 |
| S7 Customize | 190 |
Fixture Observations
- 1.The fixture provides seven stage values and three implementation totals.
- 2.The script identifies Node as fastest only because the supplied Node total is the smallest fixture value.
- 3.No measured performance comparison or feature-parity conclusion is supported by the attached evidence.
Fixture Design
The reproduction script reads pre-filled stage and total timings from sample-data.json, selects the smallest supplied median, and writes results.json. It does not clone 50 repositories, execute Node or Python Harvest pipelines, run azd, call Azure OpenAI, or collect timing samples.
Limitations
- • No pipeline implementation is executed by the reproduction script.
- • No raw timing distribution or 50-repository corpus is attached.
- • The compared workflows are not demonstrated to be feature-equivalent.
- • The fixture cannot support a performance or superiority claim.
Replay This Fixture
Replay the deterministic sample transformation. The script, sample data, and expected output are in the repository:
git clone https://github.com/frootai/frootai-core cd frootai-core/scripts/benchmarks/pipeline-speed bash run.shView script on GitHub