More control over slice element pointers

[adam-stanek]

Hi guys, I have a question or possibly feature request for you.

I am using ent in my project. For it, it makes perfect sense to use omit_slice_element_pointers: false because it internally uses all the entities as pointers. However I have part of the code base which does not expect that and works with structs. For me it is rather painful to do the conversion for big number of arrays. Is there some elegant solution to the problem I haven't thought of?

I am going to give you an example to illustrate what I am dealing with.

I have a custom scalar AssetID. It is string composed from 2 parts, index name and document id. However it is opaque to the API. For the sake of example it can be simplified as

type AssetID struct {
  IndexName string
  DocumentID string
}

func (aid AssetID) String() string { /* ... */}

Additionally I have MarshalAssetID(aid AssetID) graphql.Marshaler and UnmarshalAssetID(v interface{}) (AssetID, error) defined for the scalar.

In the schema whenever I use it in schema [AssetID!] i get []*AssetID which I then need to translate to []AssetID because that is what I use in the rest of the application. I have already quite extensive usage in the code base before I stumbled upon this problem and don't want to refactor it - there is no real benefit to having it as a pointer, quite the opposite.

Is there a way, how to force a particular field to be omitting the pointer (and produce []AssetID over []*AssetID in this case)?

I think it would be awesome to have it configurable on field by field basis. Or am I overlooking something?

Thank you

[boojongmin]

#2178 same discussion.