[DDC-1628] onUpdate parameter on @JoinColumn not supported Created: 31/Jan/12  Updated: 07/Jan/14  Resolved: 21/Mar/13

Status: Closed
Project: Doctrine 2 - ORM
Component/s: Mapping Drivers
Affects Version/s: 2.2
Fix Version/s: None

Type: Task Priority: Major
Reporter: Chris Richard Assignee: Marco Pivetta
Resolution: Invalid Votes: 3
Labels: None


 Description   

It looks like this is in the older documentation (2.0) but not mentioned in the latest.

I need to use ON UPDATE CASCADE in a few cases. Seems odd to support onDelete and not onUpdate. Am I missing something?



 Comments   
Comment by Benjamin Eberlei [ 31/Jan/12 ]

We removed onUpdate because we couldnt come up with a use-case. CAn you share yours?

Comment by Chris Richard [ 08/Mar/12 ]

I assume it's just the mysql driver but all the constraints get created such that you cannot change the primary keys (even if it's not an auto-gen PK). ON UPDATE CASCADE would probably be a better default.

Comment by Kaspars Sproģis [ 28/Aug/12 ]

I really hope onUpdate annotation attribute will be restored.
I used it in PostgreSQL very often. In some entities in some projects Primary Key ID can be very important its sequence, and if you want to change it, then "ON UPDATE CASCADE" changed it for all the references too. It was must have feature. Now few of my applications are broken with latest doctrine orm.
Please consider restoring it back.
Thank you

Comment by Václav Novotný [ 11/Oct/12 ]

+1 - onUpdate attribute is very useful for PostgreSQL databases, it controls operations with foreign keys.

Comment by Kenneth Kataiwa [ 10/Nov/12 ]

What do you mean, you couldn't come up with a use-case? If one is migrating data from one table to another and requires that all foreign key references be changed to map the ones in the new table created, before deleting the last table. And there are may more case. Maybe I am missing some thing here, if it's a use-case for the MySQL and PostgreSQL guys, why would you not support it.

Comment by Marco Pivetta [ 10/Nov/12 ]

Kenneth Kataiwa updating identifiers is something you don't do, and I sincerely also don't have a use case for this, not in the context of an ORM at least. Also, handling changes in your identifiers in the UnitOfWork is a real problem that adds a lot overhead.

Comment by Václav Novotný [ 12/Nov/12 ]

+1 - onUpdate attribute is very useful for PostgreSQL databases, it controls operations with foreign keys.

I'm taking my words back. It is not a good idea to define onUpdate on foreign keys. Yes, PostgreSQL supports it but it doesn't mean that we should use it in ORM.
Thanks for your time.

Comment by Marco Pivetta [ 21/Mar/13 ]

This issue is invalid, since data migration is not a task for the ORM anyway.

Comment by Gary Golden [ 01/May/13 ]

I have a third party table which holds users:

CREATE TABLE `user`
(
`user_id` INT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY,
`email` VARCHAR(255) NOT NULL UNIQUE,
) ENGINE=InnoDB CHARSET="utf8";

In my table I want to use natural foreign key, so I reference `email`.

CREATE TABLE `transaction` (
`id` INT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY,
`user_email` VARCHAR(255) NOT NULL,
FOREIGN KEY (`user_email`) REFERENCES `user`(`email`)
);

I would like to RDBMS handle email updates on the foreign records.

That is a real-life use case of the onDelete, which you decided to remove.
Please, get it back if possible.

Comment by I. S. [ 07/Jan/14 ]

I have a good use case and I am really missing the onUpdate cascade.
However it is working inside a many-to-one association but not in a one-to-one association.
The use case is a little complicated but I can send it to you if it could change something.

Generated at Mon Oct 20 11:33:12 UTC 2014 using JIRA 6.2.3#6260-sha1:63ef1d6dac3f4f4d7db4c1effd405ba38ccdc558.