#3374 text_field() does not use overridden model accessors - Ruby on Rails

Type	To find
responsible:me	tickets assigned to you
tagged:"@high"	tickets tagged @high
milestone:next	tickets in the upcoming milestone
state:invalid	tickets with the state invalid
created:"last week"	tickets created last week
sort:number, importance, updated	tickets sorted by #, importance or updated
Combine keywords for powerful searching.
Use advanced searching »

This project is archived and is in readonly mode.

#3374 ✓committed

text_field() does not use overridden model accessors

Reported by Dennis Theisen | October 14th, 2009 @ 02:23 AM | in 3.0.2

The following is an old problem, which still hasn't been fixed.

I couldn't find the same issue on Lighthouse, so I submitted the ticket again. If you this is a duplicate please feel free to delete it, but I can confirm the problem still exists in 2.3.4 and it feels like it should be fixed! (there is even a patch already)

This was the original ticket on Trac: http://dev.rubyonrails.org/ticket/3519

When utilizing text_field, the value of the text field when the form is populated from the database is not being pulled from the overridden Model accessor, but is instead being directly pulled from the Model. Example:

In the View: <%= text_field( 'db', 'name' ) %>

In the Model:

def name
  # Return the name of the db, less the prepend of 'sitename_' name = read_attribute(:name) site = self.site return name = name.gsub(site.name + "_",)
end

This overridden method works great when calling db.name.

Expected result: text_field uses overridden method like everyone else. Actual result: text_field just grabs the direct data.

Comments and changes to this ticket

You flagged this item as spam.
Dennis Theisen October 14th, 2009 @ 02:24 AM
- Tag changed from “text_field” to “form_for, text_field”
Rohit Arondekar June 15th, 2010 @ 02:31 PM
- Assigned user set to “Rohit Arondekar”
Dennis, can you confirm if this issue still exists on 2.3.8 and Rails master?
Rohit Arondekar June 16th, 2010 @ 03:01 AM
- Tag changed from “form_for, text_field” to “accessor, form_for, text_field”
- State changed from “new” to “open”
- Assigned user changed from “Rohit Arondekar” to “Jeremy Kemper”
I can confirm this issue exists on Rails master.

The model has a title field.
```
class Post < ActiveRecord::Base
  def title
    "blah blah"
  end
end
```
The edit form that should show "blah blah" in the text field.
```
<%= form_for(@post) do |f| %>
  <div class="field">
    <%= f.label :title %><br />
    <%= f.text_field :title %>
  </div>
  <div class="actions">
    <%= f.submit %>
  </div>
<% end %>
```
It fetches the text directly from the DB bypassing the title method.
Rohit Arondekar June 16th, 2010 @ 03:02 AM
Can somebody please confirm on the 2-3-stable branch or 2.3.8?
You flagged this item as spam.
Sam Soffes June 16th, 2010 @ 03:04 AM
- Tag changed from “accessor, form_for, text_field” to “form_for, text_field”
- Assigned user changed from “Jeremy Kemper” to “Rohit Arondekar”
I just tested this in Rails 3.0.0.beta4 and experienced the same problem.

I created a new app, ran rails g migration post title:string, and made post look like this:
```
class Post < ActiveRecord::Base
  def title
    "blah blah"
  end
end
```
The textfield in both the new and edit, didn't show "blah blah" as the title, but an empty string or what was in the database in edit action. In the console, it behaved as expected.
Rohit Arondekar June 16th, 2010 @ 03:10 AM
- Assigned user changed from “Rohit Arondekar” to “Jeremy Kemper”
Santiago Pastorino June 18th, 2010 @ 11:02 PM
- no changes were found...
Santiago Pastorino June 18th, 2010 @ 11:04 PM
- Milestone cleared.
- Assigned user changed from “Jeremy Kemper” to “Santiago Pastorino”
Santiago Pastorino June 18th, 2010 @ 11:10 PM
- Milestone cleared.
You flagged this item as spam.
Miles Egan July 16th, 2010 @ 06:38 AM
- Importance changed from “” to “Low”
This seems to be a result of this definition in lib/action_view/form_helper.rb:
```
        def value_before_type_cast(object, method_name)
          unless object.nil?
            object.respond_to?(method_name + "_before_type_cast") ?
            object.send(method_name + "_before_type_cast") :
            object.send(method_name)
          end
        end
```
Which means that we only call the native method if we can't find a corresponding before_type_cast version. This seems backwards. If we define an accessor that should take priority, right? Reversing the sense of this method to call method_name first and then fall back to before_type_cast doesn't seem to break any tests.
Repository August 1st, 2010 @ 11:41 PM
- State changed from “open” to “committed”
(from [fb0bd8c1092db51888ec4bb72af6c595e13c31fa]) Makes form_helper use overriden model accessors

[#3374 state:committed] http://github.com/rails/rails/commit/fb0bd8c1092db51888ec4bb72af6c5...
Repository August 1st, 2010 @ 11:58 PM
(from [8141f0894ecf869ee11f077e5b49e06669790632]) Makes form_helper use overriden model accessors backport

[#3374] http://github.com/rails/rails/commit/8141f0894ecf869ee11f077e5b49e0...
You flagged this item as spam.
Chris Wise August 26th, 2010 @ 03:00 PM
With the commit to RC2, there are some side-effects to this with Time fields and MySQL (2.8.1 gem). It used to be that when text_field was used with a Time field (not Date), that the value would be rendered as "13:00:00" (as an example). Now it renders as "Sat Jan 01 13:00:00 UTC 2000".

Is this more of a problem with the MySQL gem, or in ActiveRecord (I am sorry, I can't follow the code well enough to know)? Also in the committed fix (http://github.com/rails/rails/commit/fb0bd8c1092db51888ec4bb72af6c5..., why would the object.send(method_name + "_before_type_cast") be allowed w/o checking if the object respond to it (Again, I am weak at working through the code and trying to figure out if that method is required to be there for each attribute - maybe it is).
```
   def value_before_type_cast(object, method_name)
     unless object.nil?
       object.respond_to?(method_name) ?
       object.send(method_name) :
       object.send(method_name + "_before_type_cast")
     end
   end
```
You flagged this item as spam.
Chris Wise August 26th, 2010 @ 03:34 PM
I rescind my comment. I was informed I was using mysql gem (2.8.1) and should've been using mysql2 gem instead. There is some funny behaviour from the mysql gem where I was getting an expected rendering as a time but that's apparently not what the behaviour is expected to be.
You flagged this item as spam.
Oliver Eilhard August 26th, 2010 @ 05:56 PM
The commit 8141f0894ecf869ee11f077e5b49e06669790632 seems to break a User in one of my applications that has a write-only password field.

The User model comes with a password attribute which is write-only and hashed before written to the DB, so there's only the following code in the User model:
```
  def password=(cleartext_password)
    @password = cleartext_password
  end
```
In the form I'm using the form helper like this:
```
  f.password_field :password
```
Now when rendering the form, the form_helper is complaining about a missing password_before_type_cast method which it didn't before RC2.
Repository August 27th, 2010 @ 01:08 PM
(from [f95ba5c262bfd713f0a2fc656f8e645d3eea60f2]) Make InstanceTagMethods#value_before_type_cast raise if the model don't respond to attr_before_type_cast or attr method

[#3374] [#5471 state:committed] http://github.com/rails/rails/commit/f95ba5c262bfd713f0a2fc656f8e64...
Repository August 27th, 2010 @ 01:09 PM
(from [3ba8e3100548f10fce0c9784981a4589531476dd]) Make InstanceTagMethods#value_before_type_cast raise if the model don't respond to attr_before_type_cast or attr method

[#3374] [#5471 state:committed] http://github.com/rails/rails/commit/3ba8e3100548f10fce0c9784981a45...
You flagged this item as spam.
Noah September 22nd, 2010 @ 03:20 AM
This commit makes the error message a bit obscure. If, for example, you use a custom Person object and present the field "name", but your Person object doesn't define name I would expect to see "undefined method name' for #<Person:0xf661188>" and not "undefined methodname_before_type_cast' for #<Person:0xf661188>"
Santiago Pastorino September 24th, 2010 @ 09:49 PM
Right send a patch file ;)
You flagged this item as spam.
Tom Stuart September 28th, 2010 @ 11:37 AM
There's some complaining on the 2.3 backport, specifically about the fact that this breaks the behaviour whereby the user's original inputs (vs their typecasted equivalents) are kept on the form after a validation failure.

No longer a problem for 2.3.x since it's been reverted on 2-3-stable, but out of interest, why's this not a problem for Rails 3 too?
Santiago Pastorino September 28th, 2010 @ 05:40 PM
Tom we discussed about this and i'm going to revert it from Rails 3 too.
You flagged this item as spam.
Tom Stuart September 29th, 2010 @ 09:07 AM
Great, Santiago. Thanks for clarifying!
You flagged this item as spam.
Tom Stuart October 11th, 2010 @ 06:32 AM
Resolved by this commit, I believe. Thanks.
Jeremy Kemper October 15th, 2010 @ 11:01 PM
- Milestone set to “3.0.2”
[bulk edit]