chrisdutz opened a new pull request, #19175:
URL: https://github.com/apache/iotdb/pull/19175

   ## Description
   
   Both SQL parsers try the faster SLL prediction mode first and fall back to 
full LL on a syntax error, and the LL stage reports the error. The SLL stage 
still used:
   
   - the default error strategy (tree model), or
   - the reporting strategy (table model).
   
   Both run `sync()` at every rule and attempt error recovery, whose result is 
thrown away anyway.
   
   The changes:
   
   - `StatementGenerator`: the SLL parser uses `BailErrorStrategy`.
   - `SqlParser` (table model): the SLL stage uses `BailErrorStrategy`; before 
the LL stage the parser switches back to the existing reporting strategy. The 
SLL stage now also catches the `ParseCancellationException` that the bail 
strategy throws.
   
   Parse results and error messages are unchanged. I compared 28 tree and 24 
table statements, valid and invalid; `SqlParserErrorHandlerTest` and 
`StatementGeneratorTest` pass.
   
   ## Performance
   
   SQL inserts on a standalone node, mean of several runs with a 95 % 
confidence interval:
   
   | Model | Throughput | Server CPU per value |
   |---|---|---|
   | Tree | +6.4 % [+2.0, +10.7] | -6.3 % |
   | Table | +16.3 % [+11.2, +21.5] | -13.5 % |
   
   ## Tests
   
   398 tests in the parser and relational analyzer packages ran. One fails: 
`ExtractExpressionTest.constantFoldTest`, which fails the same way on master 
without this change.
   
   🤖 Generated with [Claude Code](https://claude.com/claude-code)
   


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to