fields_for nested attributes

[fredwu]

Hi,

I noticed the following code snippet in lib/signed_form/form_builder:

def fields_for(record_name, record_object = nil, fields_options = {}, &block)
  hash  = {}
  array = []

  if nested_attributes_association?(record_name)
    hash["#{record_name}_attributes"] = fields_options[:signed_attributes_context] = array
  else
    hash[record_name] = fields_options[:signed_attributes_context] = array
  end

  add_signed_fields hash

  content = super
  array.uniq!
  content
end

It seems the hash["#{record_name}_attributes"] is simply an empty array. I assume this is intentional - but it also means that in our form (which by the way could be an issue in our form since we are not dealing with strictly ActiveRecord objects), blah_attributes will still get rejected because the array is empty.

In our form we have to manually call add_signed_fields with :blah_attributes => [:a, :b, :c] to get them accepted.

Is this by design? Any thoughts?

[fredwu]

After further debugging, I nailed it down to being an issue caused by simple_fields_for. If I use fields_for, everything works, but using simple_fields_for doesn't. SignedForm::FormBuilder doesn't get initialised so the FIELDS_TO_SIGN loop never gets triggered for the nested fields.

I'll dig further.

[tubbo]

This seems to be the same issue as #17. Does it only crop up when using SimpleForm?

[ready4god2513]

The issue on #17 that I am finding is that signed_form doesn't allow for the override of the scalar value. If you are using non AR values (hashes, arrays, etc) signed_form will mark the fields as strings, but arrays can get sent in (and rejected).

[Spaceghost]

I might need some help looking into this. @tubbo, do you have any context on this?