apache / apache/arrow-java

DenseUnionVector is not transferred correctly

未关闭
#200 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
Type: bug
主要语言
Java
星标
94
派生
152
平均合并
3 天 16 小时
30 天内合并 PR
11

描述

### Describe the bug, including details regarding any error messages, version, and platform.

Hello,

I encountered a weird issue when converting a `Table` to a `VectorSchemaRoot`. All the `DenseUnionVector` were empty after the conversion.
From my understanding, it looks like there is a bug in the `TransferImpl.transfer` method of the `DenseUnionVector`.
After copying the `valueCount` value to the second vector [in this line](https://github.com/apache/arrow/blob/45918a90a6ca1cf3fd67c256a7d6a240249e555a/java/vector/src/main/codegen/templates/DenseUnionVector.java#L634), it should also set the `nextTypeId`.
After the transfer, `nextTypeId` is equals to `0`, so trying to transfer again would create an empty vector as [the for loop is done against the `nextTypeId`](https://github.com/apache/arrow/blob/45918a90a6ca1cf3fd67c256a7d6a240249e555a/java/vector/src/main/codegen/templates/DenseUnionVector.java#L628)

Proposed fix would be as simple as :

```java
@Override
public void transfer() {

...

for (int i = 0; i < nextTypeId; i++) {
if (internalTransferPairs[i] != null) {
internalTransferPairs[i].transfer();
to.childVectors[i] = internalTransferPairs[i].getTo();
}
}
to.valueCount = valueCount;
to.nextTypeId = nextTypeId;
clear();
}
```

I will raise a PR with the change.

### Component(s)

Java

贡献指南

打开贡献指南

调研方向

从 java/vector/src/main/codegen/templates/DenseUnionVector.java 开始,重点查看 TransferImpl.transfer 以及使用 nextTypeId 的循环。复现 Table 到 VectorSchemaRoot 的转换,并验证传输后的 DenseUnionVector 是否保留其值计数和类型状态,从而确保后续传输不会为空。

由索引模型根据 Issue 内容生成。

评估

技术栈
java
领域
data
Issue 类型
缺陷
难度
2/5
预计耗时
1-3 小时
活跃度
停滞
描述清晰度
描述清楚
新手友好度
50/100

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。