You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@texera.apache.org by GitBox <gi...@apache.org> on 2026/07/06 18:12:03 UTC

[GH] (texera/backport/python-int-null-schema-coercion-v1.2): Workflow run "Required Checks" failed!

The GitHub Actions job "Required Checks" on texera.git/backport/python-int-null-schema-coercion-v1.2 has failed.
Run started by GitHub user eugenegujing (triggered by eugenegujing).

Head commit for run:
a28a313c1ced7b751021c22b2f4850ee53562fac / eugenegujing <eu...@outlook.com>
fix(python-worker): coerce integral floats to int for INT/LONG fields

Backport of #6053 to release/v1.2.

A numpy-backed int column cannot hold NaN, so pandas promotes an INT/LONG
column containing nulls to float64 (119 -> 119.0). The Python worker's
strict schema validation in Tuple.finalize() then crashed with an
"Unmatched type ... expected AttributeType.INT, got 119.0" error.

Tuple.cast_to_schema() now coerces such integral floats back to int for
INT/LONG fields, but only within a safe range (INT: Arrow int32 capacity;
LONG: the float64 exact-integer window [-(2^53)+1, 2^53-1]); non-integral,
infinite, and out-of-range floats are left unchanged so validation still
fails loudly. INTEGRAL_TYPE_RANGES lives next to the other per-AttributeType
maps in models/schema/attribute_type.py.

Fixes #5935

Report URL: https://github.com/apache/texera/actions/runs/28812993826

With regards,
GitHub Actions via GitBox