ddl: Allow create timestamp default 0 without sql_mode NO_ZERO_DATE (#30305)

[mjonss]

What problem does this PR solve?

Issue Number: close #30305

Problem Summary:
create table t (ts timestamp default 0) was not allowed if sql_mode did not include NO_ZERO_DATE

What is changed and how it works?

Allow to create and alter it.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

None

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has been approved by:

• bb7133
• tangenta

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[bb7133]

This change looks good to me, but the CI failed:

[2021-12-08T00:31:27.730Z] FAIL: column_type_change_test.go:2153: testColumnTypeChangeSuite.TestCastFromZeroIntToTimeError
[2021-12-08T00:31:27.730Z] 
[2021-12-08T00:31:27.730Z] column_type_change_test.go:2190:
[2021-12-08T00:31:27.730Z]     tk.MustGetErrCode("alter table t modify column a date;", mysql.ErrTruncatedWrongValue)
[2021-12-08T00:31:27.730Z] /home/jenkins/agent/workspace/tidb_ghpr_check_2/go/src/github.com/pingcap/tidb/util/testkit/testkit.go:402:
[2021-12-08T00:31:27.730Z]     tk.c.Assert(err, check.NotNil)
[2021-12-08T00:31:27.730Z] ... value = nil

#30305 is a regression introduced by #25728, @tangenta please give some suggestions on the fix.

[tangenta]

MySQL's behavior of column-type-change is controlled by the "strict" mode. Even NO_ZERO_DATE is specified, the date can be zero...

set @@sql_mode = 'STRICT_TRANS_TABLES';
drop table if exists t;
create table t (a int);
insert into t values (0);
alter table t modify column a date;
-- ERROR 1292 (22007): Incorrect date value: '0' for column 'a' at row 1


set @@sql_mode = 'NO_ZERO_DATE';
drop table if exists t;
create table t (a int);
insert into t values (0);
alter table t modify column a date;
-- Query OK, 1 row affected, 1 warning (0.11 sec)
-- Records: 1  Duplicates: 0  Warnings: 1

[tangenta]

tidb/ddl/backfilling.go

Lines 614 to 616 in 2844660

	sessCtx.GetSessionVars().StmtCtx.AllowInvalidDate = sqlMode.HasAllowInvalidDatesMode()
	sessCtx.GetSessionVars().StmtCtx.DividedByZeroAsWarning = !sqlMode.HasStrictMode()
	sessCtx.GetSessionVars().StmtCtx.IgnoreZeroInDate = !sqlMode.HasStrictMode() || sqlMode.HasAllowInvalidDatesMode()

Adding

sessCtx.GetSessionVars().StmtCtx.NoZeroDate = sqlMode.HasStrictMode()

here should fix the test.

[sre-bot]

Code Coverage Details: https://codecov.io/github/pingcap/tidb/commit/f3e1640558f46f1cef83bb85ec5e304f94faba07

[mjonss]

/cc @bb7133 @tangenta

[dianqihanwangzi]

/run-check_dev_2

[bb7133]

LGTM

[tangenta]

/merge

[ti-chi-bot]

This pull request has been accepted and is ready to merge.

Commit hash: 1c830fc

[bb7133]

/merge

[bb7133]

/run-check_dev_2

[ti-chi-bot]

@mjonss: Your PR was out of date, I have automatically updated it for you.

At the same time I will also trigger all tests for you:

/run-all-tests

If the CI test fails, you just re-trigger the test that failed and the bot will merge the PR for you after the CI passes.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository.