This project is archived and is in readonly mode.

Nested forms reject_if is behaving stupidly

Activity

Eloy Duran
- Assigned user set to Eloy Duran
- State changed from new to invalid
accepts_nested_attributes_for :phone_numbers, :allow_destroy => true, :reject_if => proc { |attr| attr[:number].blank? }

Now it doesn't because the attribute needs to be a string and not a symbol

accepts_nested_attributes_for :phone_numbers, :allow_destroy => true, :reject_if => proc { |attr| attr["number"].blank? }

Are you giving it a hash yourself? For instance, from a test case?

If that's the case, you need to choose one form or the other, not both. In a regular request flow you will always get a HashWithIndifferentAccess. So it won't matter.

This should work both ways.

I disagree. This same “issue” comes up in other parts of Rails as well, but most of the times it's in test cases. You as the developer should use the appropriate form.

It would also be nice to support a method on the reject_if call accepts_nested_attributes_for :phone_numbers, :allow_destroy => true, :reject_if => :blank? Which would call the phone_number.blank? method This was the way it worked with the attribute_fu plugin which was the best nested forms implementation in Rails before this came along

I would happy to apply such a patch.

But please open a new ticket for it. As the issue you have, which is mentioned in the title, is a totally different one.

March 2nd, 2009 @ 11:14 AM
david.cizek (at gmail)
The first issue: attr[:number].blank? or attr["number"].blank? cause problems for me, too. Not using testcase, just regular behaviour. From debugging (attrs are the ones pushed to Proc):
```
(rdb:1) attrs
{"value"=>""}
(rdb:1) attrs[:value]
nil
(rdb:1) attrs[:value].blank?
true
```
Little more testing:
```
(rdb:1) attrs
{"value"=>"text"}
(rdb:1) attrs.class
Hash
(rdb:1) a=HashWithIndifferentAccess.new
{}
(rdb:1) a.class
HashWithIndifferentAccess
```
May 21st, 2009 @ 03:16 PM

Eloy Duran

Hi David,

I'm not quite sure what it is that you're saying. Could you please elaborate your point? Thanks.

For what it's worth, this is what I get from your examples which work for me as expected:

% irb -rrubygems -ractivesupport
irb(main):001:0> h = HashWithIndifferentAccess.new
=> {}
irb(main):002:0> h["value"] = ""
=> ""
irb(main):003:0> h
=> {"value"=>""}
irb(main):004:0> h["value"].blank?
=> true
irb(main):005:0> h[:value].blank?
=> true
irb(main):006:0> h["value"] = "text"
=> "text"
irb(main):007:0> h["value"].blank?
=> false
irb(main):008:0> h[:value].blank?
=> false

May 21st, 2009 @ 04:58 PM

david.cizek (at gmail)
Hi Eloy,
sorry for my poor articulation.

The problem seems to be that in the example

accepts_nested_attributes_for :phone_numbers, :allow_destroy => true, :reject_if => proc { |attr| attr[:number].blank? }

the attr seems NOT to be a HashWithIndifferentAccess but just Hash and therefore the Proc condition always fail as written (even if there is nested non blank attr["number"]), needs to be

...:reject_if => proc { |attr| attr["number"].blank? }

and then it works.
David

May 28th, 2009 @ 10:39 AM