Allow unit return type for suspend functions

[oguzcelik]

Summary

This pr aims to fix the lint rule for retrofit usage. Solves this issue

For suspend functions it's a valid case to return Unit.

Requirements (place an x in each [ ])

Unfortunately the contributing guidelines and Contributor license agreement links are not working.

• I've read and understood the Contributing Guidelines and have done my best effort to follow them.
• I've read and agree to the Code of Conduct.

The following point can be removed after setting up CI (such as Travis) with coverage reports (such as Codecov)

• I've written tests to cover the new code and functionality included in this PR.

The following point can be removed after setting up a CLA reporting tool such as cla-assistant.io

• I've read, agree to, and signed the Contributor License Agreement (CLA).

[salesforce-cla]

Thanks for the contribution! Before we can merge this, we need @oguzcelik to sign the Salesforce Inc. Contributor License Agreement.

[ZacSweers]

Let's make this more intentional. Could we either add an opt-in annotation like @AllowUnitResponse with a doc on it explaining contexts where it makes sense? Blanket disabling this feels like the wrong move when the goal is to make developers more intentional.

[oguzcelik]

I followed other examples of how to make the test recognise this import. Don't know if there is a better way.

[oguzcelik]

Marked as Binary retention as we don't need this at runtime.

[ZacSweers]

Can it just be SOURCE?

[oguzcelik]

Would you want to make the docs more explicit ? Something like:

/**
 * Annotation to allow suspend Retrofit functions to return Unit.
 * 
 * This annotation is used to indicate that a suspend Retrofit function is allowed to have 
 * a return type of [Unit]. This might be necessary for certain HTTP methods where the 
 * response body is not significant. For instance:
 * 
 * - Methods like `@POST`, `@PUT`, or `@DELETE` may logically return `Unit` if 
 *    the client is only concerned with the success/failure of the operation and not the 
 *    actual response body.
 * 
 * Invalid Use Cases:
 * - `@GET`: Typically, `@GET` methods should return a meaningful response object as 
 *    they are intended to retrieve data.
 */

[ZacSweers]

Sorry for the delay, I was on sabbatical over the summer. Left a couple comments, also please run ./gradlew spotlessApply to fix formatting issues (and in the future - please match our project's conventions)

[ZacSweers]

Can it just be SOURCE?

[ZacSweers]

If this is false we should be able to skip needing the isVoidOrUnitReturnType variable right? Let's make that logic lazier

[ZacSweers]

Sorry @oguzcelik but you've not signed our CLA, so I cannot accept this PR or finish it for you. Instructions were commented here.

[oguzcelik]

Hey @ZacSweers 🖖 hope you enjoyed your sabbatical 🥳

I had already mentioned in the pr description that the CLA link unfortunately doesn't work.

Here's a video in case you are wondering what I mean

Screen.Recording.2024-09-03.at.10.55.58.mov

Please let me know once the cla link and the contributing guidelines links are fixed and I'll try to get back to this to open another pr 👍

[ZacSweers]

Please use the one in the comment I linked above? The template is unfortunately old but the comment from the salesforce bot should be working