You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@shardingsphere.apache.org by "lzcGeek (via GitHub)" <gi...@apache.org> on 2026/09/25 00:00:46 UTC
[I] [Bug] Firebird rowsClause (ROWS m TO n) does not resolve its operands and maps the inclusive end row as a row count [shardingsphere]
lzcGeek opened a new issue, #39956:
URL: https://github.com/apache/shardingsphere/issues/39956
## Bug Report
**For English only**, other languages will not accept.
Before report a bug, make sure you have:
- Searched open and closed [GitHub issues](https://github.com/apache/shardingsphere/issues)
- Read documentation: [ShardingSphere Doc](https://shardingsphere.apache.org/document/current/en/overview/)
Please pay attention on issues you submitted, because we maybe need more details.
If no response anymore and we cannot reproduce it on current information, we will **close it**.
Please answer these questions before submitting your issue. Thanks!
### Which version of ShardingSphere did you use?
master @ `22137b6`, with the `ROWS` support for `limitClause` in place (see the note at the end — the branch this is based on is the one proposed in #39955, without which the clause is not visited at all and this path is unreachable).
### Which project did you use? ShardingSphere-JDBC or ShardingSphere-Proxy?
Both, though this is a parser-level defect and is visible without a database.
### Expected behaviour
Firebird's `ROWS m TO n` form selects the inclusive row range `m .. n`, so it should parse to a `LimitSegment` with `offset = m` and `rowCount = n - m + 1`. For `SELECT * FROM t_order ROWS 100000 TO 100009` that is `offset = 100000`, `rowCount = 10`.
### Actual behaviour
The `rowsClause` branch of `FirebirdDMLStatementVisitor.visitLimitClause` does not resolve its operands — both come out `null`:
```
SELECT * FROM t_order ROWS 100000 TO 100009
-> limit=[22,42] offset=null rowCount=null
SELECT * FROM t_order ROWS 100000
-> offset=null rowCount=null
SELECT * FROM t_order ORDER BY order_id ROWS 5
-> offset=null rowCount=null
```
### Reason analyse (If you can)
The branch passes `ctx.rowsClause().expr(i)` to the `visitLimitValue` helper:
```java
if (null != ctx.rowsClause()) {
offset = (PaginationValueSegment) visitLimitValue(ctx.rowsClause().expr(0));
if (ctx.rowsClause().expr().size() > 1) {
rowCount = (PaginationValueSegment) visitLimitValue(ctx.rowsClause().expr(1));
}
}
```
`visitLimitValue` only recognises contexts whose single child is a `NumberLiteralsContext` or a `ParameterMarkerContext` — that is `limitOffset` and `limitRowCount`, the sub-rule contexts used by the `OFFSET` / `FETCH` path. The single child of an `expr` is a `booleanPrimary`, so neither branch matches and the helper falls through to `null`, discarding both values while still building a `LimitSegment` with anchors.
Independently of the lookup, the mapping is wrong: `rowsClause` is `ROWS expr (TO expr)?` and Firebird's range is inclusive, so the end value is a row *number*, not a row count. `offset = m, rowCount = n - m + 1` is the conversion; the current code would need to compute that once the operands are readable.
### Steps to reproduce the behaviour, such as: SQL to execute, sharding rule configuration, when exception occur etc.
Parse any of the three statements above with the Firebird dialect. No sharding rule is needed — the failure is visible in the parsed statement: `LimitSegment` is present but `getOffset()` and `getRowCount()` are both empty.
### Example codes for reproduce this issue
The standalone probe used to produce the output above parses the statement with `FirebirdParserFacade` and prints the `LimitSegment` anchors and values; happy to attach it if useful.
---
Note on scope: this is filed separately on purpose. #39926 / #39955 cover the `OFFSET ... ROWS FETCH FIRST ... ROWS ONLY` path, which is what the rewrite rule needed and what was reported; no existing `sql-case` covers the `ROWS m TO n` grammar, so that change was kept out of the PR. I am willing to submit a PR for this issue (add the operand lookup for `expr`, apply the inclusive-range conversion, and add `sql-case` entries with expectations for both `ROWS m` and `ROWS m TO n`).
I have searched the open and closed issues for `ROWS TO` / `rowsClause` and found no existing report.
Disclosure: found while preparing #39955 with an AI coding assistant (DeepSeek Harness), which was used to read the Firebird grammar and visitor and to run the local parser probe.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@shardingsphere.apache.org.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org