executor: support prepare DDL statements with no parameters

[ian-p-cooke]

What problem does this PR solve?

This PR improves compatibility with MySQL 5.7 by accepting some DDL as prepared statements.
Currently TiDB accepts no DDL as prepared statements (#9425).
MySQL 5.7 lists some DDL statements in the accepted SQL for prepared statements: https://dev.mysql.com/doc/refman/5.7/en/sql-syntax-prepared-statements.html

What is changed and how it works?

I removed the explicit check for DDL and relaxed the restriction from no DDL to ddl without parameters. I don't believe parameters are used often for DDL and I am unable to test DDL with parameters so I left that restriction in. Even so it's still useful; for example, Diesel's MySQL driver prepares all statements, including DDL, and with this change TiDB becomes more compatible with MySQL 5.7 and Diesel.

Check List

Tests

• Unit test in prepared_test.go TestPreparedNameResolver

Code changes

no function signatures have changed.
The text of ErrPrepareDDL changed.

Side effects

• SQL that was illegal before becomes legal (DDL with no parameters in prepared statements).

Related changes

• Need to be included in the release note

[CLAassistant]

All committers have signed the CLA.

[ngaut]

Thank you @ian-p-cooke

I can't figure out why we need to prepare DDL statements, could you explain your user case?

[ian-p-cooke]

@ngaut the case I ran into was using Diesel's "migrate" feature with the mysql driver against a TiDB instance. The driver assumes that all SQL statements that it sends can be prepared, including the DDL. This works fine for a mysql instance but a TiDB instance fails on the DDL.

I talked to Diesel's maintainer and he suggested that TiDB would be better supported with a custom TiDB driver but I don't have time to work on that and the mysql driver works except for this DDL issue.

Since this issue improves the compatibility with mysql I thought it would be a good change to TiDB.

[ian-p-cooke]

I think I messed up my pull request be rebasing on master manually. can this pull request be repaired or should I just start a new one?

[codecov]

Codecov Report

Merging #10144 into master will increase coverage by 0.0151%.
The diff coverage is 0%.

@@               Coverage Diff               @@
##            master     #10144        +/-   ##
===============================================
+ Coverage   77.673%   77.6882%   +0.0151%     
===============================================
  Files          411        411                
  Lines        85435      85408        -27     
===============================================
- Hits         66360      66352         -8     
+ Misses       14110      14097        -13     
+ Partials      4965       4959         -6

[shenli]

@ian-p-cooke I think you can reset your branch and use force-push to update your repo.

[ngaut]

Thanks for detail explanation @ian-p-cooke

[ian-p-cooke]

ah, i used the 'update branch' button instead this time and now the pull request is back down to just my commits. I think you can squash them you pull, right? I won't do anything else with git right now.

[winkyao]

LGTM

[winkyao]

@crazycs520 @bb7133 PTAL

[ngaut]

LGTM