# Is this dry? (models & migrations) **URL:** \ **Category:** Rails\ **Created:** [August 31, 2007, 4:25pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486 "2007-08-31T16:25:58Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [August 31, 2007, 4:25pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/1 "2007-08-31T16:25:58Z")
I’m creating a tool for a preexisting distributed SQL database. One of the things I’m doing is taking an SQL dump of the schema and transforming it into models. It strikes me that there appears to be a lot of duplication between the model and the database. – Unique keys are defined both in the model (validates\_uniqueness\_of) and in the database (ALTER TABLE [dbo].[data] ADD CONSTRAINT UNIQUE ( [name], [deleteDate] ) ON [PRIMARY]). – Relations are defined in both the model (belongs\_to) and in the database (ALTER TABLE [dbo].[data] ADD CONSTRAINT FOREIGN KEY ( [checkOutSuerID] ) REFERENCES [dbo].[users] ( [userID]) NOT FOR REPLICATION) – Null allowances are defined both in the model (validates\_exclusion\_of :column, :in =\> [nil]) and in the database ( [column] [varchar] (21) NOT NULL) This doesn’t seem very dry. I’m wondering what’s going on. Also, I’m wondering if I should be adding field size limits to my models as well. If so, does this also go for fields which are the default size ([varchar] (255))? ---
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [August 31, 2007, 4:51pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/2 "2007-08-31T16:51:01Z")
On 8/31/07, Student [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > ( [checkOutSuerID] ) REFERENCES [dbo].[users] ( [userID]) NOT FOR > REPLICATION) > – Null allowances are defined both in the model > (validates\_exclusion\_of :column, :in =\> [nil]) and in the database > ( [column] [varchar] (21) NOT NULL) > > This doesn’t seem very dry. I’m wondering what’s going on. There’s some discussion about this in the Agile Web D. With Rails book. Some people prefer to include integrity constraints in the database as well as in Rails, and some people will rely solely on Rails to maintain the integrity of the database by model validations. The latter has the advantage of being database agnostic as well as keeping all of your logic where it should be - in the model. > Also, I’m wondering if I should be adding field size limits to my > models as well. If so, does this also go for fields which are the > default size ([varchar] (255))? it depends… Do you want users to be able to enter as much information into a field as they want, only to have the result (unknowingly to them) truncated? Or do you want to enforce a constraint on the number of letters they enter, receiving an error message if they go past this value? Adam ---
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [August 31, 2007, 4:57pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/3 "2007-08-31T16:57:02Z")
I’ve only gone through the book once, and this is my first project, so I must have forgotten/missed what was there. The index didn’t help much ☹ I am wondering about whether or not I need the model to enforce this information or if I can rely on the adapter to throw when the db refuses to do a save. (Note that in my case, the db already has the information.) It’s not clear to me that there is much of a difference. You seem to be implying that an attempt to write an overly-large string will silently truncate in the db? (Uggh) Does RoR require an explicit size check when the size is the default? Of course, if this were the development list, I would be wondering why RoR cannot read all of this stuff from the db & stick it into the model the same way that it does the column names & types. 🙂 Nathan ---
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [September 1, 2007, 4:21pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/4 "2007-09-01T16:21:14Z")
Rails is of the opinion that constraints such as these belong in the application layer. While you can certainly use database-level constraints to enforce your business rules at the level of the database (and some would argue that you should in any case), rails will work best if you specify them in the model. Trying to generate human readable errors from exceptions thrown from constraints in the database, for instance, doesn’t work very well in practice so I would suggest that you continue to use your validations and other application-level constraints, DRY or not. Rein ---
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [August 31, 2007, 6:18pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/5 "2007-08-31T18:18:11Z")
On 8/31/07, Student [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > I’ve only gone through the book once, and this is my first project, so > I must have forgotten/missed what was there. The index didn’t help > much ☹ now that I have the book in front of me, there isn’t a large section discussing this, it was actually in a small footnote on page 141 where Dave talks about some developers not bothering with database defined foreign key constraints while he personally prefers to use them. > I am wondering about whether or not I need the model to enforce this > information or if I can rely on the adapter to throw when the db > refuses to do a save. (Note that in my case, the db already has the > information.) If you rely only on the database to throw an exception when a constraint is violated, how will you filter this down to your rails app? You’ll have to make sure you catch any exceptions (which of course you should be doing anyways), determine what exception was raised and then display the appropriate error. Wouldn’t it be easier to simply use rails’ built in validations to handle this, and use error\_messages\_for to display the messages to the user? (and have the database constraints in place to offer you further protection in the event that you write some bad code) > It’s not clear to me that there is much of a > difference. like I mentioned in my previous post, using Rails to enforce data integrity is database agnostic and it keeps all your logic in the model. If another programmer were to work on your application, they’d be able to quickly see all the validations directly from within Rails, whereas if you relied solely on the database, they’d have to go searching in another place to find this information. Plus, you gain a bunch of convenience methods when using Rails’ built in validations. > You seem to be implying that an attempt to write an > overly-large string will silently truncate in the db? (Uggh) Does > RoR require an explicit size check when the size is the default? yes, the data will be silently truncated and rails does need an explicit size check to prevent this from happening (or at least to prevent the record from being saved if the field is over a certain number of characters) > Of course, if this were the development list, I would be wondering why > RoR cannot read all of this stuff from the db & stick it into the > model the same way that it does the column names & types. 🙂 Here’s a plugin which may be of use to you: [http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/](http://drnicwilliams.com/2006/08/07/ann-dr-nics-magic-models/) Adam ---
**Author:** ![student](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/ecc23a/32.png) [@student](https://www.ruby-forum.com/u/student)\ **Post date:** [September 1, 2007, 5:39pm UTC](https://www.ruby-forum.com/t/is-this-dry-models-migrations/113486/6 "2007-09-01T17:39:24Z")
rein.henrichs wrote: > Rails is of the opinion that constraints such as these belong in the > application layer. Rails is of the opinion that you can add (and make) plugins to synchronize database and program validations. DrySQL leaps to mind… And note that each database constraint is typically a unique index, so this might optimize all the excessive re-queries that ActiveRecord runs on all the constrained tables at .save time… – Phlip [http://www.oreilly.com/catalog/9780596510657/](http://www.oreilly.com/catalog/9780596510657/) “Test Driven Ajax (on Rails)” assert\_xpath, assert\_javascript, & assert\_ajax