Bump policyengine-core floor to 3.24.4 (closes #8066)#8069
Closed
MaxGhenis wants to merge 1 commit intoPolicyEngine:mainfrom
Closed
Bump policyengine-core floor to 3.24.4 (closes #8066)#8069MaxGhenis wants to merge 1 commit intoPolicyEngine:mainfrom
MaxGhenis wants to merge 1 commit intoPolicyEngine:mainfrom
Conversation
Core 3.24.4 ships the fixes from PolicyEngine/policyengine-core#475: - `Simulation.apply_reform` preserves `set_input` values (the H3 cache invalidation used to wipe user-provided dataset inputs) - `Holder.get_array` walks up `simulation.parent_branch` before falling back to `default`, so nested branches (e.g. `no_salt` under `itemizing`) still see ancestor inputs - `GroupPopulation.clone` passes the cloned population to holder.clone so branch-aware lookups resolve correctly Together these unblock the strict breakdown/children validator added in core 3.24.0. Before 3.24.4, bumping the floor triggered `TypeError: int() argument ... not 'NoneType'` in the `tax_unit_itemizes.yaml` integration test (state_fips was wiped by apply_reform) plus an infinite recursion in `tax_liability_if_itemizing`. With 3.24.4 (or later), the existing `test_system_import.py` regression test now actually catches breakdown/children mismatches at CI time — the class of bug that caused issue PolicyEngine#8055 and the three partial-fix patches PolicyEngine#8045, PolicyEngine#8049, PolicyEngine#8051. Verified locally against core 3.25.0: all 794 IRS tests pass. Also remove the `[tool.uv]` environments restriction (`python_version >= '3.11'`) since core 3.24+ supports 3.9/3.10 directly on PyPI — the lockfile now resolves cleanly across the full 3.9-3.14 range. Closes PolicyEngine#8066. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2 tasks
Contributor
Author
|
Superseded by a rebased PR (the original branch accumulated 3 rounds of GitHub runner preemptions / test cancellations; rebasing onto current main gives a clean CI run). |
This was referenced Apr 18, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Bump
policyengine-corefloor to 3.24.4 and drop the temporary Python-version restriction onuv.lock. Closes #8066.Why
policyengine-core 3.24.0added a strict validator that errors onbreakdown/ parameter-children mismatches — the exact class of bug that hit Martin Holmer on us 1.644.0 and 1.645.0 (issue #8055) and that three partial-fix PRs had to chase: #8045, #8049, #8051.But bumping the core floor used to be blocked by three coupled regressions inside core 3.24.0–3.24.3:
Simulation.apply_reformwiped user-provided dataset inputs (the H3 cache invalidation hit_memory_storage._arraysfor user inputs as well as formula-output caches).Holder.get_arrayfell back only todefault, missing the parent-branch walk needed by country-package nested branches (e.g.no_saltunderitemizing).GroupPopulation.clonepassed the source population (not the clone) toholder.clone, so group-entity holders on cloned branches pointed at the wrong simulation.Together these caused the
tax_unit_itemizes.yamlintegration test to crash withTypeError: int() argument ... not 'NoneType'(state_fips wiped) plus an infinite recursion intax_liability_if_itemizing.PolicyEngine/policyengine-core#475 fixed all three, shipped in core 3.24.4 (latest at time of writing: 3.25.0).
Effect
test_system_import.py(added in Add regression test for parameter breakdown mismatch (closes #8055) #8058) now actually does work: constructingCountryTaxBenefitSystem()runs the strict validator, so any future breakdown/children mismatch fails CI rather than reaching end users at import time.[tool.uv]environments = ["python_version >= '3.11'"]restriction can be removed — core 3.24+ supports 3.9 and 3.10 on PyPI.Verified locally
policyengine_us/tests/policy/baseline/gov/irs/with core 3.25.0.tax_unit_itemizes.yaml7/7 pass (was 2 failing on core 3.24.1).test_system_import.pypasses as expected.Test plan