Column analyses errors results in falling back to string types
- Dominant language
- Java
- Stars
- 14.1k
- Forks
- 3.8k
- Avg merge
- 2d 58m
- Merged PRs (30d)
- 233
Description
This is related to, but has a different impact than https://github.com/apache/druid/issues/18437.
[18437](https://github.com/apache/druid/issues/18437) and its [fix](https://github.com/apache/druid/pull/19162) focused on row signature flapping, which can cause queries to use incorrect columns at query planning time when column analysis errors occur.
This issue, however, focuses on type correctness problems that arise when column analysis errors cause a fallback to string types.
The root cause of this behavior is still unclear. There is some discussion in https://github.com/apache/druid/pull/19162#discussion_r2941563639.
It will be good to detect these types more accurately rather than fallback to string type when there are column analyses errors.
### Affected Version
32.0.1 (likely affects later versions as well)
Contributor guide
Research direction
Review issue 18437, its fix in pull request 19162, and the linked discussion to understand how column-analysis errors reach string-type fallback. The issue does not name files or tests; done means detecting affected column types accurately instead of falling back to string types.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- databases
- Issue type
- Bug
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100