Using a hash as the data object does not seem to work. about rabl

[nesquena]

I never considered using a hash as the object, interesting idea. I always used a class (ORM) record as the object. I guess I don't see a reason a Hash couldn't be made to work but not surprising it doesn't. There are a few assumptions I made about the object right now and one of them is that the object responds to the "valid?" method and that it has dot notations to access attributes.

Here's how I would achieve what you want for now:

object false
node(:foo) { "bar" }
node(:pop) { 123 }

or if its a nested hash:

object false
node(:thing) { @thing }

I have to give this some thought, this is a use case I never considered supporting.

[nesquena]

Also as for the undefined methodto_html' for #Rabl::Engine:0x1075c0730` it looks like you are trying to use RABL during an HTML request (not a json or xml request). You probably shoudn't allow the action to respond to HTML or have an alternate erb or haml template for that case...

[kimptoc]

Hi,

Thanks for the feedback - will try what you suggest.

I tried creating a local object like this:

class Thing
    attr_accessor :foo, :pop
    def initialize(a,b)
      @foo =a
      @pop =b
    end

  end

So the controller did this:

@thing2 = Thing.new("boo",987)

And template is this:

object @thing2
attributes :foo, :pop

Getting 'null' response (when I specify the format).

On the to_html stuff, yesterday I got away without it - not sure how. Wonder if there is a way to make json the default format.

Thanks,
Chris

[kimptoc]

Hi,

Just to say your suggestion with the "object false" does the job - and is clearer in my controller, so I prefer it - the hash was a contrived object to get the data to the view - so better to not have it :)

Thanks,
Chris

[nesquena]

@kimptoc the to_html error comes from this change yesterday: ed9ec55 in which I detect the request format if theres no explicit format. I will have to fix it by disallowing html as the format I think and default to json.

As for the Thing record, try this for fun:

class Thing < Struct.new(:foo, :pop)
   def initialize(h)
     super *h.values_at(*self.class.members.map {|s| s.intern})
   end

  def valid?; true; end
end

and

@thing2 = Thing.new(:foo => "boo", :pop => 987)

Can you tell me if that works?

[nesquena]

@kimptoc Okay, I am actually going to close this for now, I have never had the need to use a hash as the object. I don't think I will in the future either. In the case of constructing a response from scratch I would go with the object false approach :) If you or someone else finds this issue, I am not against the idea if someone else would like to patch it in with a pull request. I will fix the "to_html" issue tho soon. Thanks!

[kimptoc]

Hi,

The Thing suggestion works just fine and yes, please close it - was going to suggest that.

Regards,
Chris

[nesquena]

@kimptoc Fixed the 'to_html' issue here, can you confirm? 7f97b7e

[kimptoc]

Hey @nesquena - perfect, works like a charm :)

Many thanks.

[nesquena]

Excellent thanks man, pushing out a small gem point release.