Michael Milton

Filter across a join without using a filter function

2

It would be nice to be able to do something like: ```python class UserFilter(FilterSet): class Meta: model = User fields = { 'group.name': [...], } ``` Which would: * Allow...

Allow additional URL parameters to HyperlinkRelated

1

This is basically a reiteration of #54. I have an API where a report conceptually has multiple samples. Thus, the sample URL has two parameters, e.g. (my example uses Flask-Restful):...

Allow relationships to be compared to scalars

2

This is based on [the recommendations part of the spec](https://jsonapi.org/recommendations/#filtering), which lets you do things like: `GET /comments?filter[post]=1`, even though `filter[post]` isn't actually an integer. In addition, I've allowed queries...

Allow providing `get_schema_kwargs` as a function

1

Allows users to provide fields like `get_schema_kwargs` as functions instead of dictionaries. These functions have access to the view's args and kwargs. For example: ```python class PersonDetail(ResourceDetail): schema = person_schema...

Assign correct endpoint to each resource instance

2

By default we assign each Resource class with a view/endpoint. However if the same resource is used for multiple routes, the endpoint will be overwritten. e.g. previously this test would...

Custom content negotiation

1

* Allow an API to define a dictionary of converter functions for different content types, that are run before and after the JSON API schema sees the data. For example:...

I realise this has been discussed briefly at #13. However, we don't yet have a good solution. The problem is this: I want my API to input and output data...

Validate that the ID field exists, but don't load it

2

Following on from discussion here: https://github.com/json-api/json-api/issues/1427. Basically, it makes sense for JSON API servers to validate that there is an ID in all data it deals with, BUT there are...

Closes #263. One thing I'd like feedback on is the actual interface here. You could argue that `include_data=True` isn't elegant, so I'd be open to something like `include_data='all'` or `include_data='*'`.

If I have a series of nested models (from SQLAlchemy, for example), it would be nice to be able to `RootSchema(include_data=True).dump(root_model)`. Currently, you have to spell out each and every...