apache / apache/datafusion-sqlparser-rs
Should the `generic` dialect support numeric literals with underscore separators?
- Dominant language
- Rust
- Stars
- 3.5k
- Forks
- 772
- Avg merge
- 4d 9h
- Merged PRs (30d)
- 17
Description
Right now, the `generic` dialect (DataFusion's default) does not support numeric literals with separators:
```sql
DataFusion CLI v54.0.0
> select 1_000;
+------+
| _000 |
+------+
| 1 |
+------+
```
I'm wondering if the `generic` dialect should support this, just like Postgres/DuckDB/SQLite/Clickhouse, so DataFusion can use them directly without needing to switch to another dialect.
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by reproducing `select 1_000;` in the DataFusion CLI and trace how the `generic` dialect tokenizes and parses numeric literals. Compare its behavior with the Postgres, DuckDB, SQLite, or ClickHouse dialects mentioned in the issue; done means the generic dialect accepts numeric separators without treating `_000` as a column name.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust, sql
- Domain
- databases
- Issue type
- Feature
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 65/100