Modifying the column type(needn't to change the data) involved in the generated column is not completely prohibited

[zimulala]

Bug Report

Please answer these questions before submitting your issue. Thanks!

There is currently no support for changing the type of a column involving generating a column. Related to compatibility-issues-with-mysql.

1. Minimal reproduce step (Required)

 create table t2(id int, a int, b int generated always as (abs(a)) virtual);
 alter table t2 modify column a bigint;

2. What did you expect to see? (Required)

Return an error like Unsupported modify column: tidb_enable_change_column_type is true, oldCol is a dependent column 'a' for generated column.

3. What did you see instead (Required)

Query OK, 0 rows affected (2.52 sec)

4. What is your TiDB version? (Required)

Release Version: v4.0.0-beta.2-2732-gd9e2bd3da
Edition: Community
Git Commit Hash: d9e2bd3da49ec3e9a062831fb6e4ab1219a8d158
Git Branch: master
UTC Build Time: 2021-04-27 06:27:55

[zimulala]

Related to #24310.

[wjhuang2016]

I think it's ok to allow it. The document should be refined

[zimulala]

The reason for the ban here is that there is no energy to carry out the complete test at present. Before the complete test is completed, it can be prohibited by the same treatment as the modification of data(#24310).

[wjhuang2016]

If generate column is GA, then you can't forbid it. If not, then we don't need to forbid it.

[zimulala]

But we said we don't support this behavior in https://docs.pingcap.com/tidb/stable/generated-columns#limitations. So I think we need to forbid it.
In addition, if a feature is not fully supported, that is, it is available to users, but the results are uncertain, I think it is better to disable it.

[wjhuang2016]

But we said we don't support this behavior in https://docs.pingcap.com/tidb/stable/generated-columns#limitations. So I think we need to forbid it.
In addition, if a feature is not fully supported, that is, it is available to users, but the results are uncertain, I think it is better to disable it.

Which statement?

[AilinKid]

/cc @zimulala

[tisonkun]

@wjhuang2016 @zimulala I'm unsure how this issue tracked by at least this line https://github.com/pingcap/tidb/pull/25263/files#diff-0485729b654feaebae798e651571d5fd1091fbedb374ac61518e1536de7f37c4R161 shows that the issue is resolved?