[FabianHofmann]

Closes #1054.

Note

The following content was generated by AI.

Changes proposed in this Pull Request

LinearExpression.to_constraint("==", rhs) kept the sign "==" instead of the canonical "=". As a result, repr() of the constraint raised KeyError: '==' and the wrong sign was stored in the model. On current master, this happens on both the dense and the sparse backend.

The cause: signs were only normalized when add_constraints received a separate sign argument. Constructing a Constraint itself never normalized them.

• Constraint._bind now normalizes the sign, so every construction path (to_constraint, from_dataset, adding a constraint to a model) stores canonical signs.
• maybe_replace_signs is vectorized because it now runs on every constraint construction. It only converts the distinct non-canonical values and still raises ValueError on invalid signs.
• A new allow_null flag lets _bind keep NaN signs on absent rows, which shift/reindex produce. User-facing paths in model.py keep the strict default, so a NaN sign passed by the user still raises.
• New tests cover "==", "<" and ">" through to_constraint on dense and sparse (sign, repr, stored sign), and through Constraint.from_dataset.

Checklist

• AI-generated content is marked (see AGENTS.md).
• Code changes are sufficiently documented; i.e. new functions contain docstrings and further explanations may be given in doc.
• Unit tests for new features were added (if applicable).
• A note for the release notes doc/release_notes.rst of the upcoming release is included.
• I consent to the release of this PR's code under the MIT license.

[codspeed]

Merging this PR will not alter performance

✅ 181 untouched benchmarks
⏩ 181 skipped benchmarks¹

---

_{Comparing fix/1054-to-constraint-sign (d069964) with master (185c001)}

Footnotes

1. 181 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩