linuxpi opened a new issue, #26072:
URL: https://github.com/apache/datafusion/issues/26072
### Describe the bug
`array_min` / `array_max` on a zero-row `List<Utf8>` (or any non-primitive
element type) fails:
```
Execution error: Empty iterator passed to ScalarValue::iter_to_array
```
A zero-row `List<Int64>` works. `array_min_max_helper`
(`datafusion/functions-nested/src/min_max.rs`) only has a fast path for
primitive types; the fallback builds a `Vec<ScalarValue>` and calls
`ScalarValue::iter_to_array`, which rejects an empty iterator.
Zero-row batches do reach these functions in practice: `ProjectionExec` and
`TopK` evaluate expressions on every batch, so `ORDER BY array_min(tags) LIMIT
n` fails whenever an input emits an empty batch.
### To Reproduce
```rust
use std::sync::Arc;
use datafusion::arrow::array::{ArrayRef, ListBuilder, StringBuilder};
use datafusion::arrow::record_batch::RecordBatch;
use datafusion::prelude::SessionContext;
#[tokio::main]
async fn main() -> datafusion::error::Result<()> {
let ctx = SessionContext::new();
let tags: ArrayRef =
Arc::new(ListBuilder::new(StringBuilder::new()).finish());
ctx.register_batch("t", RecordBatch::try_from_iter(vec![("tags",
tags)])?)?;
ctx.sql("SELECT array_min(tags) FROM t").await?.show().await?; // error
Ok(())
}
```
Seen on 54.0.0; the code is unchanged on `main`.
### Expected behavior
Zero rows, the same as for primitive element types.
### Additional context
Found in OpenSearch's DataFusion backend when sorting on multi-valued
keyword fields with an empty shard. Proposed fix: return
`new_empty_array(&array.value_type())` for an empty input before the fallback.
PR to follow.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
