mirror of
https://github.com/langchain-ai/datafusion.git
synced 2026-07-21 03:05:28 -04:00
kosiew
4a41587bdf
## Which issue does this PR close? * [Comment](https://github.com/apache/datafusion/pull/20202#discussion_r2804841561) on #20202 --- ## Rationale for this change The `custom_file_casts` example defines a *logical/table* schema that uses `id: Int32` as the target type. In practice, casting and projection paths in DataFusion can produce **nulls** (e.g. failed casts, missing values, or intermediate expressions), and examples should avoid implying that nulls are impossible when demonstrating casting behavior. Marking the `id` field as **nullable** makes the example more realistic and prevents confusion when users follow or adapt the example to scenarios where nulls may appear. --- ## What changes are included in this PR? * Update the logical/table schema in `custom_file_casts.rs` to define `id` as **nullable** (`Field::new("id", DataType::Int32, true)`). * Adjust the inline comment to reflect the nullable schema. --- ## Are these changes tested? No new tests were added. This is a documentation/example-only change that updates a schema definition and comment. The example continues to compile and can be exercised by running the `custom_file_casts` example as before. --- ## Are there any user-facing changes? Yes (example behavior/expectations): * The `custom_file_casts` example now documents `id` as nullable, aligning the example schema with situations where cast/projection may yield null values. * No public APIs are changed and no breaking behavior is introduced.