MySQL with Soft-Deletion, Unique Key and Foreign Key Constraints

Viewed 10789

Say I have two tables, user and comment. They have table definitions that look like this:

CREATE TABLE `user` (
  `id`       INTEGER NOT NULL AUTO_INCREMENT,
  `username` VARCHAR(255) NOT NULL,
  `deleted`  TINYINT(1) NOT NULL DEFAULT 0,
  PRIMARY KEY (`id`),
  UNIQUE KEY (`username`)
) ENGINE=InnoDB;
CREATE TABLE `comment` (
  `id`      INTEGER NOT NULL AUTO_INCREMENT,
  `user_id` INTEGER NOT NULL,
  `comment` TEXT,
  `deleted` TINYINT(1) NOT NULL DEFAULT 0,
  PRIMARY KEY (`id`),
  CONSTRAINT `fk_comment_user_id` FOREIGN KEY (`user_id`)
    REFERENCES `user` (`id`)
    ON DELETE CASCADE
    ON UPDATE CASCADE
) ENGINE=InnoDB;

This is great for enforcing data integrity and all that, but I want to be able to "delete" a user and keep all its comments (for reference's sake).

To this end, I've added deleted so that I can SET deleted = 1 on a record. By listing everything with deleted = 0 by default, I can hide away all the deleted records until I need them.

So far so good.

The problem comes when:

  • A user signs up with a username (say, "Sam"),
  • I soft-delete that user (for unrelated reasons), and
  • Someone else comes along to sign up as Sam, and suddenly we've violated the UNIQUE constraint on user.

I want users to be able to edit their own usernames, so I shouldn't make username the primary key, and we'll still have the same problem when deleting users.

Any thoughts?

Edit for clarification: Added following RedFilter's answer and comments below.

I'm concerned with the case where the "deleted" users and comments are not visible to the public, but are visible only administrators, or are kept for the purpose of calculating statistics.

This question is a thought experiment, with the user and comment tables just being examples. Still, username wasn't the best one to use; RedFilter makes valid points about user identity, particularly when the records are presented in a public context.

Regarding "Why isn't username the primary key?": this is just an example, but if I apply this to a real problem I'll be needing to work within the constraints of an existing system that assumes the existence of a surrogate primary key.

5 Answers

My practical solution for soft-delete is archiving by creating a new table with the following columns: original_id, table_name, payload, (and an optional primary key `id).

Where original_id is the original id of deleted record, table_name is the table name of the deleted record ("user" in your case), payload is JSON-stringified string from all columns of the deleted record.

I also suggest making an index on the column original_id for latter data retrievement.

By this way of archiving data. You will have these advantages

  • Keep track of all data in history
  • Have only one place to archive records from any table, regardless of the deleted record's table structure
  • No worry of the unique index in the original table
  • No worry of checking foreign index in the original table

The is already a discussion here explaining why soft-deletion is not a good idea in practice. Soft-delete introduces some potential troubles in the future such as counting records, ...

You can do the following:

 CREATE UNIQUE INDEX YourIndexName 
 ON YourTableName (Column1, Column2, Column3)
 WHERE IsDeleted = 0

You can find the complete answer here:

enter link description here

Related