apache / apache/arrow-java

Avro adapter: setPosition doesn't work with BaseVariableWidthVectors

Open
#833 2 comments 0 reactions 0 assignees View on GitHub
Type: bug
Dominant language
Java
Stars
94
Forks
152
Avg merge
3d 16h
Merged PRs (30d)
11

Description

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

`BaseVariableWidthVector`-based vectors (`VarBinaryVector`, `VarCharVector`) expect the data to be written sequentially. This means `Consumer.setPosition` is not safe to use. Imagine the following scenario:

- write index 0
- write index 2
- write index 1
- write index 3

On the last step, `BaseVariableWidthVector.fillHoles` erases the data written for the index 2:

```java
protected final void fillHoles(int index) {
for (int i = lastSet + 1; i < index; i++) {
setBytes(i, emptyByteArray, 0, emptyByteArray.length);
}
lastSet = index - 1;
}
```

Here, index would be 3, lastSet + 1 would be 2 (1 + 1), so `setBytes(2, emptyByteArray, ...)` is called.

Not sure what the right solution is here, but at a minimum, the library shouldn't allow calling `setPosition` for Consumers using `BaseVariableWidthVector` vectors.

Contributor guide

Open the contributing guide

Research direction

Start by tracing the Avro adapter's Consumer.setPosition behavior with BaseVariableWidthVector-based VarBinaryVector and VarCharVector instances. Reproduce the write sequence 0, 2, 1, 3 and add a regression test; done means positioning no longer erases previously written data or is prevented for these vectors.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
data
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.