hello wezel
This repository exercises Wezel's GitHub Actions runner integration.
The line-count experiment records generated LLVM IR lines.
The intentionally small program provides a stable baseline for comparison.
Runs are assigned by Fiflok and executed by the repository workflow.
The workflow checks out the exact commit assigned to each run.
Results are uploaded to Wezel after the experiment completes.
The fixture is intentionally dependency-free for fast, deterministic runs.
Commit history is kept linear so regression attribution is easy to inspect.
This project is test data rather than a production application.
The second attribution trial is applied to main with a fast-forward.
Its commits remain individually reachable from the tracked branch.
Wezel can therefore enqueue midpoint measurements during its hunt.
Only one commit in the sequence changes the measured workload.
Later commits deliberately leave the measurement unchanged.
That makes the expected culprit unambiguous.
The trial exercises automatic dispatch rather than manual queueing.
Its expected endpoint is a Discord culprit notification.
The fixture can be removed after the end-to-end test completes.