You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Sergey Kalashnikov (JIRA)" <ji...@apache.org> on 2017/11/08 15:03:00 UTC
[jira] [Commented] (IGNITE-6276) SQL: Investigate parser generators
[ https://issues.apache.org/jira/browse/IGNITE-6276?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16244102#comment-16244102 ]
Sergey Kalashnikov commented on IGNITE-6276:
--------------------------------------------
We have identified several shortcomings with the generated parser approach.
- Error messages generated by parser are not customizable enough to the point when user may easily understand them. I doubt a user will understand messages like 'no viable alternative'.
- The ANTLR lexer cannot be controlled by parser making many useful things impossible.
- The performance assesment results aren't really great in terms of scalability.
> SQL: Investigate parser generators
> ----------------------------------
>
> Key: IGNITE-6276
> URL: https://issues.apache.org/jira/browse/IGNITE-6276
> Project: Ignite
> Issue Type: Improvement
> Components: sql
> Reporter: Sergey Kalashnikov
> Assignee: Sergey Kalashnikov
> Fix For: 2.4
>
> Attachments: antlr4-ignite.zip
>
>
> Now ignite relies on H2 for SQL processing. It has been discussed many times on dev list that we must start introducing our own SQL core in small incremental steps.
> Let's start with analyzing the options for implementing the parser part.
> We may begin with http://www.antlr.org/ and create a simple separate project that would generate the parser for some simple DDL commands like DROP INDEX.
> This will give us a hint on the complexity and limitations of the approach.
> 1) Set up Maven/ANTLR.
> 2) Prepare lexer/parser.
> 3) Generate.
> 4) Write a test.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)