4paradigm / 4paradigm/OpenMLDB

Inconsistent results in online/offline mode while join condition includes null

Open
#3,685 2 comments 0 reactions 2 assignees Claimed by @aceforeverd View on GitHub
bug execute-engine high-priority
Dominant language
C++
Stars
1.7k
Forks
331
Avg merge
12d 12h
Merged PRs (30d)
1

Description

```
table: t1
columns: [ "c1 string","c2 int","c3 bigint","c4 timestamp" ]
indexs: [ "index1:c1:c4" ]
rows:
- [ NULL,2,3,1590738989000 ]
```
```
table: t2
- columns: [ "c1 string","c2 int","c3 bigint","c4 timestamp" ]
indexs: [ "index1:c1:c4" ]
rows:
- [ NULL,2,3,1590738989000 ]
```
`sql: select t1.c1,t1.c2,t2.c3,t2.c4 from t1 last join t2 ORDER BY t2.c4 on t1.c1=t2.c1;`
expect rows:
- [ null,2,3,1590738989000 ]

real rows:
- [ null,2,null,null ]

Contributor guide

Open the contributing guide

Research direction

The issue shows a SQL join query with a NULL equality condition producing different results in online vs offline modes. Start by locating the join implementation for 'last join' in the query engine, likely in the SQL execution or optimizer modules. Examine how NULL comparisons are handled in join conditions across execution modes. Run the provided test case to reproduce the inconsistency, then trace through the join logic to identify the divergence.

Written by the indexing model from the issue text.

Assessment

Tech stack
sql
Domain
databases, machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.