Sideloading associations

[NullVoxPopuli]

Hey all, since this is a highly requested feature, I figured I'd take a stab at it since I need it for my own ember work.

Thoughts on the implementation?

The goal of this is to address needs of #938 and #899

Usage

#config/initializers/active_model_serializers.rb
ActiveModel::Serializer.config.adapter = :json
ActiveModel::Serializer.config.sideload_associations = true

And that's it. :-)
No need to have embed: :ids on all the associations.

[ZackMattor]

👍 This is working for me so far in my application! I had to tweak one thing though. I have a has_one and this sideloads it as a singleton object as apposed to an array. Before I'm pretty sure it still stuck it in an array.

Had to do the following hack...

  has_many :foos

  def foos
    [ object.foo ]
  end

VS

  has_one :foo

[NullVoxPopuli]

@criten could you elaborate? I haven't used a has_one in my app yet, so I'm uncertain what the exact issue you found is / was

[ZackMattor]

Just edited my comment.

[NullVoxPopuli]

interesting. Does has_one work in normal AMS, nested the data under foo?

[ZackMattor]

Yeah.. just checked. To be a bit more clear....

# 0.9.X
class FooSerializer < ActiveModel::Serializer
  has_one :bar
end

bar: {
  foo_id: 1
},
foos:
[{
  id: 1
}]

VS

# 0.10.X
class FooSerializer < ActiveModel::Serializer
  has_one :foo
end

bar: {
  foo_id: 1
},
foo:
{
  id: 1
}

[ZackMattor]

I think that's what its doing....

[NullVoxPopuli]

kk, I'll see if I can come up with a test and fix it

[NullVoxPopuli]

@criten I added a test but was not able to replicate. :-(

Is there anyway you have time to monkey around with the sideload_test.rb tests and see if you can replicate in there?

(I did add two more tests, though, for has_one and belongs_to)

[estum]

Heh, About a week ago I've started to make my own implementation of the same thing. Currently it is documented but doesn't contain the proper specs, that is why I haven't yet made a PR. In general It works for me, but doesn't support custom names of associations and excludes not sideloaded reflections from a "relationships" object. Also, I've added a sideloading to the SerializedResource's initialization process and now doubt about this decision.

[BenMorganIO]

👍

[ZackMattor]

I don't think this is working for me with nested has_many? Do you know if that works?

[NullVoxPopuli]

Does regular AMS work with nested has many? When I was looking through the
code, I don't think I saw any sort of recursion on the serialization bit. I
could be wrong though.

I do know that tested has many doesn't work. But I also don't know if it's
pre existing.

On Fri, Aug 28, 2015, 5:00 PM Zachary Mattor [email protected]
wrote:

I don't think this is working for me with nested has_many? Do you know if
that works?

—
Reply to this email directly or view it on GitHub
#1088 (comment)
.

[ZackMattor]

It definitely worked in version 0.9...

[NullVoxPopuli]

I'll take a look at the tests on master when I can, and see if there is a
nested serialization gap. :-)

On Fri, Aug 28, 2015, 5:06 PM Zachary Mattor [email protected]
wrote:

It definitely worked in version 0.9...

—
Reply to this email directly or view it on GitHub
#1088 (comment)
.

[NullVoxPopuli]

@criten it looks like nested has many is broken in master: NullVoxPopuli@15147de
(commit for a failing test)

[joaomdmoura]

I'd say yes, but I'm concerned with ppl using AMS outside Rails context.
I'd keep it like that and come back at it after Rails 5 release

[NullVoxPopuli]

Closing, in favor of an implementation based off my refactored work (#1111)