adds polymorphic option to association definition which includes asso… #1726
Conversation
| type: 'employee', | ||
| employee: { | ||
| id: 42, | ||
| name: 'Zoop Zoopler' |
There was a problem hiding this comment.
nested assoication
|
What's the behavior for |
|
@beauby can you elaborate what 'heterogeneous has-many' means? Im thinking nil will act the same as how it does prior, but Im wondering if at: when a relationships[association.key] is nil, that relation is skipped, but now there will be a type with a nil nesting. If you outline the desired expectations, I can write tests and code to them |
|
For the "heterogeneous has-many" case, I meant the following: |
|
@beauby ... form a pure theoretical perspective, it would act exactly as it does now, plus the effect of this PR which would add the type to the packet. such as (from the 0.9.1 docs) [http://www.rubydoc.info/gems/active_model_serializers/0.9.1] .. However, that is a big assumption that it currently acted like this. Really all this PR does was tack on one extra nest. Ill work on a test for both these as I think I know what the desired expectation is now .. thanks |
| @@ -46,7 +46,8 @@ def relationship_value_for(association, options) | |||
|
|
|||
| opts = instance_options.merge(include: @include_tree[association.key]) | |||
| relationship_value = Attributes.new(association.serializer, opts).serializable_hash(options) | |||
| if association.options[:polymorphic] | |||
|
|
|||
| if association.options[:polymorphic] && relationship_value | |||
There was a problem hiding this comment.
@beauby good call, added check for empty relationship which wont wrap the resultant in a polymorphic type if there is no association (corresponding tests added for :json and :attributes adapters)
|
Oh, right. I had just skimmed through the PR and had not understood it fully. I'm 👍 with the idea. |
|
This adds a |
|
@remear LOL, I just caught that as well, pushed it .. |
…ciation type in serializer regen gemlock regen gemlock better variable naming rubocop fixes adds to changelog adds empty relationship and has_many polymorph tests indent test cleaning -rubocop rubocop rubocop rubocop changelog remove silly .DS fix roque failure fix
Purpose
Current polymorphic association adds the association serialization directly under the polymorphic name (ie imageable). This does not allow the client provider a straightforward knowledge of what the polymorphic association is, and thus can not easily determine how to handle the data.
This PR allows for the polymorphic: option flag on the association to nest the attributes under the association type - henceforth allowing the client to know what the polymorphic association is, and handle accordingly.
Changes
adds polymorphic: option to the serializer association definition which then determines if serialized association should be nested under association type
Caveats
Affected serialization adapters: attributes && json
*Does not affect json_api *
Related GitHub issues
my earlier question/inquiry:
#1717
Earlier Polymorphic testing PR which touched up the serialization package diversion from 0.8x
#1420 (comment)
Additional helpful information
adds polymorphic option to association definition which includes association type in serializer
regen gemlock