Add decimal arithmetic kernels with checked overflow and auto-promotion + fix null mask drop on broadcasting - #185
Merged
Conversation
Decimal arithmetic uses checked operations instead of wrapping to prevent silent corruption of exact-precision values. Scale reconciliation, result-precision computation, and cross-type promotion are handled at the kernel level so callers pass DecimalArray pairs and receive correctly-typed results. Kernel operations: - Add/subtract with same-scale and cross-scale reconciliation - Multiply with result scale = s1 + s2 - Divide with numerator scale-up for fractional-digit preservation - Remainder with scale reconciliation - Negate and abs (unary, checked) - Rescale to a new scale (up via checked multiply, down via truncation) - Element-wise comparison (eq, lt, gt, le, ge, ne) with cross-scale support Dispatch routing (arithmetic_dispatch): - Same-type Decimal32/64/128 arithmetic - Cross-width promotion (Decimal32+Decimal64 widens to Decimal64, etc.) - Integer + Decimal auto-promotion (integer scaled to decimal) - Float + Decimal auto-promotion (decimal converted to Float64) - Overflow returns MinarrowError rather than wrapping - Length-1 decimal broadcast for scalar expansion Also adds KernelError::Overflow variant for checked-arithmetic failures and decimal scalar arithmetic arms for Scalar + Scalar operations.
- Fix overflow error variant and module doc wording - Add arr_dec32!, arr_dec64!, arr_dec128! macros to array.rs - Replace test helper functions with macro imports - Remove bracketed asides from comments
pbower
force-pushed
the
MA-EPC86-DECIMAL-TSK381-KERNELS
branch
from
August 31, 2026 06:07
90ea29e to
e4de091
Compare
The existing integer, float, boolean, and string broadcast arms discarded null status when repeating a length-1 array. Now all arms check the source element's null mask and broadcast it accordingly.
pbower
changed the title
pbower
added a commit
that referenced
this pull request
Aug 31, 2026
* Add DecimalArray foundation and i128 trait impls (#181) * Add Decimal32, Decimal64, Decimal128 variants to ArrowType (#180) * Wire DecimalArray into NumericArray, Array, Scalar with From impls (#182) * Wire DecimalArray into NumericArray, Array, and Scalar enums with From impls and tests (#182) * Implement decimal conversion arms in NumericArray try_ accessors * Add decimal FFI import/export with arrow-rs and Polars round-trips (#186) * Add decimal FFI import/export with arrow-rs and Polars round-trips * Integrate DecimalArray into Value, ArrayV, NumericArrayV, SuperArray, SuperArrayV, and Consolidate (#187) * Add decimal broadcast dispatch and scale-aware print formatting (#184) * Add decimal arithmetic kernels with checked overflow and auto-promotion + fix null mask drop on broadcasting (#185) * Add decimal arithmetic kernels with checked overflow and auto-promotion * Propagate null mask through broadcast for all array types * Add DecimalArray type conversions and width conversion methods (#183) * Add decimal support to pyo3 and minarrow-py Python bridges (#188) * Add Decimal32 and Decimal64 round-trip tests for arrow-rs * Add decimal support to pyo3 and minarrow-py Python bridges * Make precision and scale report values for all numeric types
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
src/kernels/arithmetic/decimal.rsmodule with checked binary ops (add, subtract, multiply, divide, remainder), unary ops (negate, abs), rescale, and comparison kernelsarithmetic_dispatchandscalar_arithmeticfor all three widthsMinarrowErrorvia newKernelError::OverflowvariantTest plan
cargo test --features decimalpasses (814 tests)cargo test --all-featurespasses (1711 tests)cargo testwithout decimal passes (no regression)