Namespace separator fix for type field in jsonapi adapter #1216

youroff · 2015-09-30T19:13:32Z

This PR removes namespace from automatically generated type field.
Discussion's going on here:
http://discuss.jsonapi.org/t/convention-for-namespacing-type/99

bf4 · 2015-09-30T22:34:35Z

test/adapter/json_api/linked_test.rb

@@ -212,7 +212,7 @@ def test_underscore_model_namespace_for_linked_resource_type
            expected = {
              related: {
                data: [{
-                  type: 'spam_unrelated_links',
+                  type: 'unrelated_links',


why did the expectation change here?

aside, the type should probably be spam:unrelated_links' orspam-unrelated_links`

json-api/json-api#850 (the issue I made that the jsonapi forum references )

(~~aside, why is there a related= writer on Post that is important?~~ just confusing naming)

youroff · 2015-10-01T04:14:17Z

The idea was to reject namespace by default:

There's no standard way to encode namespace yet.
The way how it was done before is confusing: Namespace::SomeModel was converted to namespace_some_model which is the same as it would be for the model NamespaceSomeModel.

So my idea was to cut namespace by default, but type directive will be there to set it explicitly.

youroff · 2015-10-01T04:16:43Z

For #1029 I would stick with / separator for namespace, but it's against json-api standard (hopefully it can be changed to comply that usecase). But _ or - wouldn't work as it's standard separator for words in multiword classes. Which will bring ambiguity.

bf4 · 2015-10-01T15:09:48Z

@youroff agree about _ and -. ~~how about spam:unrelated_links ? Or even spam::unrelated_links~~ ( U+003A COLON, ":" is disallowed)
per spec recommendation json-api/json-api#850 (comment) it would be spam--unrelated_links which I'm ok with

it's up to us, and we can create the ruby convention right here.

beauby · 2015-10-01T16:46:22Z

After giving it a thought, I tend to agree with @bf4. A user that uses namespaces should know what they are doing, and overriding the type with type "MyType" seems like a good option for those willing to get rid of the namespace.

youroff · 2015-10-01T17:15:39Z

Honestly, / looks more reasonable for me, cause it's common sign for paths (which namespaces in fact are for the class hierarchy), so maybe it's worth proposing for json-api team. But right now I can do -- to move it farther, if everyone agrees with this convention.

BTW, not related to this one, but I discovered recently while playing with ember, that by default it generates attribute names dasherized, and not underscored for payload. Not sure if it's explicitly stated in standard, but the point is that when we use ams with ember the keys are transferred differently encoded for inbound and outbound packets. And the payload for server doesn't seem to be ready for rails because of that, so I got a thought — are there deserializers for AR? Could AMS solve this issue be providing some wrapper for strong parameters?

beauby · 2015-10-01T17:20:36Z

@youroff Let's comply with JSON API for now. If they change the spec, we can always change AMS then.

Considering deserializers, it's a WIP, it's kinda almost ready, but we need to discuss a few things before rebasing/pushing it.

bf4 · 2015-10-01T22:56:58Z

See linked issue, ember is 'wrong'

youroff · 2015-10-01T23:45:12Z

You mean that it doesn't convert /?

bf4 · 2015-10-02T02:04:24Z

You mean that it doesn't convert /?

No, that ember uses / which is invalid

youroff · 2015-10-02T02:10:19Z

Does ember support namespaced models or it uses it somehow else?

youroff · 2015-10-02T02:11:23Z

Btw, what if we add setting for separator? So that anyone will be able to change it to / or anything else for their setup.

bf4 · 2015-10-02T02:27:03Z

legit.. new global config

bf4 · 2015-12-28T08:45:13Z

Looks like we lost sight of this. Is a good PR. @youroff Would you like to rebase and we'll finish it up?

youroff · 2016-01-14T06:01:53Z

Sorry, missed that on holidays... Sure, will take a look at it soon. What's the final decision on how we deal with namespaces?
And btw, I remember talking about global settings for deriving type name from seralizer name instead of model name. What do you think about that one? Is it worth of implementation?

bf4 · 2016-01-15T00:22:13Z

type is really the resource type, so should be derived from the object, I think.

-- is fine until we think of something better. I've been meaning to make a PR to ember to fix their impl :)

youroff · 2016-01-15T06:25:46Z

Well, in our app we use the same model in the different ways. For example, there is a User -> Membership -> Dealership relation. And Membership reflects into two different Ember models: one ProfileMembership — is a light version to be attachable to profile (and loadable with). So it's something used throughout the site to adjust current user's experience, and another Memebership is manageable by dealership Admin. It contains various additional fields, which are used on the backend. Currently I use type directive to set type explicitly inside of different serializers and having option to derive type name from the name of serializer would make the code a bit cleaner. So I believe it's not that unique a use case and having an option (with taking type from the model as a default) wouldn't hurt anyone, what do you think? And it's pretty easy to implement, I believe I've done some sketch of it already.

So for namespace, we'll have a setting for separator with '--' as default, right?

Btw, what's the way of using namespaces on the side of Ember? If you have some example in mind, could you point out?

youroff · 2016-01-15T06:36:10Z

Here's more generic example... Let's say we have Post model. It may have title, intro, body, lots of metadata. Though in a news feed we only need title and intro. So that would be PostPreview on ember's end. And when user wants to read whole article it would load Post model. Sure we could implement fields parameter, but the approach seems cleaner. Also, not sure about now, but when ember was about version 1-something, it would take record from the store by default, so once loaded reduced record, it wouldn't reload it when called with findRecord.

NullVoxPopuli · 2016-03-10T11:07:57Z

I think we need to wait for the JSON API spec to decide on what to do about namespaces, yeah?

bf4 · 2016-03-10T15:00:50Z

@NullVoxPopuli

we need to wait for the JSON API spec to decide on what to do about namespaces, yeah?

In 1.0 it is up to us. See linked issue I made re 'kebab' case in the json api repo

remear · 2016-03-17T18:17:57Z

Can we make a decision on this?

youroff · 2016-03-17T18:45:40Z

Actually, I would commit some time to fix that if we have a decission. Just ran into the need of namespace separator compatible with Ember. @bf4?

remear · 2016-03-17T20:32:23Z

I'm late to this conversation, but I'm fine with either spam--unrelated_links or spam__unrelated_links.

youroff · 2016-03-17T20:39:32Z

I think all agree that we need some global setting for json-api adapter and the last question is what would be the default.

youroff · 2016-03-20T18:02:18Z

Just created a PR #1609 for this feature. Codebase changed a lot, so I just made it from scratch. @bf4 could you review?

NullVoxPopuli · 2016-03-21T17:40:05Z

Closing in favor of #16009

type field generated without namespace when using jsonapi adapter

5a65cd2

youroff mentioned this pull request Sep 30, 2015

type directive for serializer to control type field with json-api adapter #1213

Merged

beauby added the V: 0.10.x label Sep 30, 2015

bf4 reviewed Sep 30, 2015
View reviewed changes

youroff mentioned this pull request Oct 1, 2015

Proposal for type namespaces json-api/json-api#896

Closed

Namespaces separated by '--'

9db77d8

youroff changed the title ~~Remove namespace while generating type field for jsonapi~~ Namespace separator fix for type field in jsonapi adapter Oct 1, 2015

bf4 mentioned this pull request Oct 1, 2015

Recommendation for dasherizing (kebab-case-ing) namespaced object, such as Admin::User json-api/json-api#850

Closed

bf4 added the C: JSON API label Dec 28, 2015

bf4 added the C: Feature label Jan 15, 2016

bf4 mentioned this pull request Jan 29, 2016

Document JSON API implementation defs and progress in class #1482

Merged

NullVoxPopuli closed this Mar 10, 2016

bf4 reopened this Mar 10, 2016

youroff mentioned this pull request Mar 20, 2016

Namespace separator setting #1609

Closed

NullVoxPopuli closed this Mar 21, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Namespace separator fix for type field in jsonapi adapter #1216

Namespace separator fix for type field in jsonapi adapter #1216

youroff commented Sep 30, 2015

bf4 Sep 30, 2015

youroff commented Oct 1, 2015

youroff commented Oct 1, 2015

bf4 commented Oct 1, 2015

beauby commented Oct 1, 2015

youroff commented Oct 1, 2015

beauby commented Oct 1, 2015

bf4 commented Oct 1, 2015

youroff commented Oct 1, 2015

bf4 commented Oct 2, 2015

youroff commented Oct 2, 2015

youroff commented Oct 2, 2015

bf4 commented Oct 2, 2015

bf4 commented Dec 28, 2015

youroff commented Jan 14, 2016

bf4 commented Jan 15, 2016

youroff commented Jan 15, 2016

youroff commented Jan 15, 2016

NullVoxPopuli commented Mar 10, 2016

bf4 commented Mar 10, 2016

remear commented Mar 17, 2016

youroff commented Mar 17, 2016

remear commented Mar 17, 2016

youroff commented Mar 17, 2016

youroff commented Mar 20, 2016

NullVoxPopuli commented Mar 21, 2016

Namespace separator fix for type field in jsonapi adapter #1216

Namespace separator fix for type field in jsonapi adapter #1216

Conversation

youroff commented Sep 30, 2015

bf4 Sep 30, 2015

Choose a reason for hiding this comment

youroff commented Oct 1, 2015

youroff commented Oct 1, 2015

bf4 commented Oct 1, 2015

beauby commented Oct 1, 2015

youroff commented Oct 1, 2015

beauby commented Oct 1, 2015

bf4 commented Oct 1, 2015

youroff commented Oct 1, 2015

bf4 commented Oct 2, 2015

youroff commented Oct 2, 2015

youroff commented Oct 2, 2015

bf4 commented Oct 2, 2015

bf4 commented Dec 28, 2015

youroff commented Jan 14, 2016

bf4 commented Jan 15, 2016

youroff commented Jan 15, 2016

youroff commented Jan 15, 2016

NullVoxPopuli commented Mar 10, 2016

bf4 commented Mar 10, 2016

remear commented Mar 17, 2016

youroff commented Mar 17, 2016

remear commented Mar 17, 2016

youroff commented Mar 17, 2016

youroff commented Mar 20, 2016

NullVoxPopuli commented Mar 21, 2016