Fix #452: [Model] ConsistencyOfDatabaseFrequencyTables

[GiggleLiu]

Summary

Add the implementation plan for ConsistencyOfDatabaseFrequencyTables, then execute it in follow-up commits.

Fixes #452

[GiggleLiu]

Implementation Summary

Changes

• Added the ConsistencyOfDatabaseFrequencyTables model with serializable frequency-table / known-value helpers, validation, evaluation, complexity metadata, registry exports, and focused model tests.
• Added pred create ConsistencyOfDatabaseFrequencyTables support, including compact parsers for attribute domains, published frequency tables, and known values, plus CLI coverage for direct creation and canonical examples.
• Added the ConsistencyOfDatabaseFrequencyTables -> ILP<bool> reduction using one-hot assignment variables and McCormick auxiliary variables, with closed-loop reduction tests.
• Added the paper entry for Consistency of Database Frequency Tables with the formal definition and worked example.

Deviations from Plan

• None.

Open Questions

• None.

[codecov]

Codecov Report

❌ Patch coverage is 98.73188% with 7 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (main@2fec9f9). Learn more about missing BASE report.
⚠️ Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
...s/misc/consistency_of_database_frequency_tables.rs	96.33%	7 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #718   +/-   ##
=======================================
  Coverage        ?   97.62%           
=======================================
  Files           ?      399           
  Lines           ?    49720           
  Branches        ?        0           
=======================================
  Hits            ?    48537           
  Misses          ?     1183           
  Partials        ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[GiggleLiu]

Agentic Review Report

Structural Check

Structural Review: rule ConsistencyOfDatabaseFrequencyTables -> ILP

Structural Completeness

#	Check	Status
11	Paper reduction-rule entry	FAIL — the new reduction is implemented in src/rules/consistencyofdatabasefrequencytables_ilp.rs:113 and registered for example export in src/rules/mod.rs:129, but docs/paper/reductions.typ:3097 only adds the model problem-def; there is no reduction-rule("ConsistencyOfDatabaseFrequencyTables", "ILP") block for the new edge.

Build Status

• make test: PASS
• make clippy: PASS

Semantic Review

• paper quality: ISSUE — the code, tests, and exported reduction graph now include ConsistencyOfDatabaseFrequencyTables -> ILP, but the paper does not state or justify that direct reduction, so the documentation is structurally incomplete.

Summary

• 10/11 structural checks passed
• Missing paper coverage for ConsistencyOfDatabaseFrequencyTables -> ILP; the reduction exists in code, but the paper only documents the model and omits the new reduction rule.

---

Quality Check

Quality Review

Design Principles

• DRY: OK
• KISS: OK
• HC/LC: OK

HCI (if CLI/MCP changed)

• Error messages: ISSUE — invalid CDFT frequency-table indices report “out of range for --n”, but pred create ConsistencyOfDatabaseFrequencyTables has no --n flag. The shared helper hardcodes that wording at problemreductions-cli/src/commands/create.rs:189 and the new CDFT parser reuses it at problemreductions-cli/src/commands/create.rs:219.
• Discoverability: OK
• Consistency: ISSUE — the global help correctly marks --known-values optional at problemreductions-cli/src/cli.rs:264, but the usage string injected into CDFT validation errors shows it as required at problemreductions-cli/src/commands/create.rs:1714, even though the command succeeds without that flag.
• Least surprise: OK
• Feedback: OK

Test Quality

• Naive test detection: ISSUE
• The reduction tests only use a single 2x2 table at src/unit_tests/rules/consistencyofdatabasefrequencytables_ilp.rs:11, so they never exercise the nontrivial offset logic added at src/rules/consistencyofdatabasefrequencytables_ilp.rs:21 and src/rules/consistencyofdatabasefrequencytables_ilp.rs:159 for multiple tables and non-uniform attribute domains.
• The new CLI tests only cover happy-path creation and help text at problemreductions-cli/tests/cli_tests.rs:5542 and problemreductions-cli/tests/cli_tests.rs:5586, so the broken CDFT error path above was not caught.

Issues

Critical (Must Fix)

None.

Important (Should Fix)

• The CDFT create path emits a misleading remediation hint because out-of-range attribute errors still tell users to fix --n instead of --attribute-domains / the inferred attribute count (problemreductions-cli/src/commands/create.rs:189, problemreductions-cli/src/commands/create.rs:219).
• The new ILP reduction’s key indexing logic is effectively untested: src/unit_tests/rules/consistencyofdatabasefrequencytables_ilp.rs:11 never covers multiple tables or varying domain sizes, so bugs in src/rules/consistencyofdatabasefrequencytables_ilp.rs:21 and src/rules/consistencyofdatabasefrequencytables_ilp.rs:34 would currently slip through.

Minor (Nice to Have)

• CDFT help and validation output disagree about whether --known-values is optional: the flag list marks it optional at problemreductions-cli/src/cli.rs:264, while the usage string shown on errors treats it as mandatory at problemreductions-cli/src/commands/create.rs:1714.

Summary

• Important — misleading CDFT error messages reference a nonexistent --n flag.
• Important — reduction tests miss the multi-table / non-uniform-domain cases that exercise the new indexing logic.
• Minor — CDFT usage text says --known-values is required even though help and runtime behavior treat it as optional.

---

Agentic Feature Tests

Agentic Feature Test Report

Feature: ConsistencyOfDatabaseFrequencyTables
Worktree: /Users/jinguomini/rcode/problem-reductions/.worktrees/review-pr-718
Profile: ephemeral downstream CLI user
Use case: discover the model from the CLI catalog, inspect it, create the canonical example, and solve it using docs/examples alone
Expected outcome: the full pred list → pred show → pred create --example → pred solve path works, and the explicit reduction to ILP also works
Verdict: pass with 1 confirmed setup/docs issue

Issues Found

1. Medium: README says make cli “builds target/release/pred”, but it actually performs a global cargo install and replaces the user’s existing pred.
   • Status: confirmed in current worktree
   • Evidence: README.md:40 vs Makefile:170 and Makefile:171
   • Reproduction:
     1. cd /Users/jinguomini/rcode/problem-reductions/.worktrees/review-pr-718
     2. make cli
     3. Observe cargo install --path problemreductions-cli and Replacing /Users/jinguomini/.cargo/bin/pred
   • Impact: the documented “build from source” path writes outside the repo and can overwrite an already-installed pred, which is surprising for review-pipeline and for downstream users.

Summary

Check	Result	Notes
pred list	Pass	ConsistencyOfDatabaseFrequencyTables * appears in the catalog with Rules = 1
pred show ConsistencyOfDatabaseFrequencyTables	Pass	Shows description, 4 fields, 5 size fields, and 1 outgoing reduction to ILP/bool
pred create --example ConsistencyOfDatabaseFrequencyTables	Pass	Canonical example emitted successfully
pred solve cdf.json	Pass	Solved via ILP with evaluation true
pred reduce cdf.json --to ILP	Pass	Produced a 1-step bundle ConsistencyOfDatabaseFrequencyTables -> ILP
pred solve cdf-to-ilp.json	Pass	Bundle solved successfully

Per-Feature Details

• Discoverability: good. Starting from the generic CLI docs and the catalog, the model was easy to find with pred list, and pred show exposed the reduction path clearly.
• Setup: partial. The build succeeded, but the README’s make cli description is inaccurate.
• Functionality: good. The canonical example was created and solved successfully, and the explicit reduction flow also worked.
• Expected vs actual: matched. The downstream-user scenario completed end-to-end without needing source inspection.
• Blocked steps: skipped the crates.io install path intentionally, because the goal was to test the PR worktree build, not the published crate.
• Friction points: there is no model-specific CLI example in the general CLI docs, but the generic pred create --example <PROBLEM_SPEC> guidance plus the paper example were sufficient.

Commands Executed

1. pred list
2. pred show ConsistencyOfDatabaseFrequencyTables
3. pred create --example ConsistencyOfDatabaseFrequencyTables -o cdf.json
4. pred solve cdf.json
5. pred reduce cdf.json --to ILP -o cdf-to-ilp.json
6. pred solve cdf-to-ilp.json

I did not save a report file under docs/test-reports/ because this run was requested to stay read-only.

---

Generated by review-pipeline

[GiggleLiu]

Follow-up items (recorded during final review):

• ensure_attribute_indices_in_range() in create.rs:194 hardcodes --n in error message — should parameterize the flag name for CDFT context
• CDFT usage string in create.rs:1812 shows --known-values as required, but it's optional