[alxmrs]

🪭 Claude (einfold)

The simulation, einfold's random differential test against plain DataFusion, compares results row by row after sorting them by keys and then values. It sorted values with partial_cmp, which treats -0.0 and 0.0 as equal. Where two rows have the same keys and values that differ only in a zero's sign, as when a domain repeats a key and the query fills both rows from the domain's own 0.0 and -0.0, the stable sort kept them in arrival order, which differs between einfold's run and the host's. The comparison then paired each with the other's row, and reported a sign difference that isn't one. Nothing changes for users.

This is all 14 failures of the soak in #138 (11 seeds). With values sorted by total_cmp, which orders -0.0 before 0.0, every one of the 11 seeds passes at the soak's commit. In the first, einfold ran the host's own plan (host_plans=1), so the two results were the same rows. #134's new generators are what made such domains common.

Decisions to review

• Sort by total_cmp, not by bits or as a multiset (rows). Values still pair by order within a key, so rows that agree within the float bound still pair up. The comparison of zeros themselves is unchanged: a zero must still have the host's sign.

How it is tested

The 11 seeds from #138 replayed at 8ec90e5 with this change, and the default run passes.

Closes #138.

🤖 Generated with Claude Code

https://claude.ai/code/session_01M96xSez6PZwTmfaGh2eJgS

[coderabbitai]

Important

• Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

Run configuration

• Source mode: Auto
• Repository: xqlsystems/einfold
• Branch: m1/simulation-zero-order
• File: .coderabbit.yaml
• Commit: 3c15be1
• Layers: Repository YAML
• Review profile: CHILL
• Plan: Advanced
• Run ID: 99af03aa-a793-4712-8a47-48593704a272

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[alxmrs]

🪭 Claude (einfold)

Closing in favor of #142, the same fix from the simulation's owner, opened while this one was.