[AdaWorldAPI]

Item B of the post-merge review of #1681 (GPT:gpt-ada-01/20261021-pr1674-1682-consolidated-review-01). I took it on the operator's instruction (EXO-50 /038). The original owner is session 01G22y, and the operator aligns afterwards.

Defects on main

• bool_and(TRUE, NULL) answered 0, where PostgreSQL answers TRUE.
• bool_or(NULL, NULL) answered 0, where PostgreSQL answers NULL.
• count_distinct_sorted counted a run of NULL keys by their payload.

Change (crates/lance-graph-r2il/src/procs.rs)

bool_or and bool_and. New signature: bool_or(select, valid, flag) / bool_and(select, valid, flag).

• Both fold over P = select ∧ valid.
• Each pushes the value, then present. present == 0 is SQL NULL; the value is then 0 and is not FALSE.
• bool_or = ANY(P ∧ T).
• bool_and = present − ANY(P ∧ ¬T).

count_distinct_sorted. New signature: count_distinct_sorted(key_lane, valid).

• It gates the selection on the key's presence plane before KEY_RUNS.
• An unordered lane is still refused. Nothing is sorted or hashed.

All three stay R2IL procedures interpreted by FoldDialect and lowered to existing mask-risc operations. There is no new opcode or terminal, and no crate outside r2il calls these procedures.

Evidence

New tests in tests/procs.rs:

• bool_aggregates_keep_null_apart_from_false covers the directive's six cases, each run again with every NULL payload scrambled:

  input	bool_or	bool_and
  TRUE, NULL	TRUE	TRUE
  FALSE, NULL	FALSE	FALSE
  TRUE, FALSE, NULL	TRUE	FALSE
  NULL, NULL	NULL	NULL
  empty selection	NULL	NULL
  FALSE, FALSE	FALSE	FALSE

• count_distinct_excludes_null_keys_whatever_their_payload covers the first, middle and last runs, plus an all-NULL selection.

• In the existing primitive test, the empty-selection expectation changes from (0, 1) to NULL.

Disable runs. Four, each red on its target test. Each removes valid from one place:

• bool_or;
• bool_and's FALSE fold;
• the presence fold;
• count_distinct_sorted.

PostgreSQL 16.15. Run with examples/null_aggs_pg_parity.rs, a lab instrument that prints SKIP without a server and does not count as CI parity.

• 300 queries over 60 random tables: values and NULL states are equal.
• The run includes 64 NULL bool aggregates and 12 cases of bool_and TRUE over a NULL row.
• The disable arm (validity treated as every row) differs on 224.

Gates: r2il tests pass (9 dialect + 22 procs). Clippy --all-targets -D warnings and fmt are clean.

Board: a correction section is appended to entries/2026-10-21-d-rft-1-typed-composition.md.

Merge is attended: I will not merge this PR.

---

Generated by Claude Code

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	🔄 Running since 2026-10-21T22:11:25.223891Z	4230d26	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[coderabbitai]

Important

• 🔍 Trigger review

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

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 5f024afe-7993-40e0-b4b1-03eb1d4c4533

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

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

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

_{Comment @coderabbitai help to get the list of available commands.}