docs: Fallback to reference description for responses

[Nelspike]

Checklist

• run npm run test and npm run benchmark
• tests and/or benchmarks are included
• documentation is changed or added
• commit message and code follows the Developer's Certification of Origin
  and the Code of conduct

This PR attempts to resolve the issue on #795 by resolving the $ref given in the schema and fetching that description in case there's none. PTAL! 🚀

[jean-michelet]

Thanks for the PR!

I don't know the code base so I can't approve myself.

[jean-michelet]

Does it solve your bug @melroy89?

[Nelspike]

First of all, @jean-michelet, thank you for the review! 🚀

Secondly and equally thankful to @melroy89 for the detailed reproduction description 💪🏼 From what I can see from your code, the only reason why this wouldn't work is due to the trailing # on your $ref, which is notFoundError#. That $ref is absolutely correct, so we need to find a way to resolve these properly in this PR. If you want to give it a try without the trailing #, I think that would do the trick, at least for this example!

[Nelspike]

@melroy89 - That's exactly what I meant in my reply, we need to consider those use cases because your reference is 100% correct 😅 The code I pushed in the PR didn't take that into account yet, but just pushed a change that does! I never meant that you should remove #, but that the code I had previously worked without it.

[Nelspike]

Seems like the second use case you mentioned on your EDIT is blocked by this, @melroy89: #676

At the moment, seems like definitions is not supported within OpenAPI definitions on fastify-swagger, which difficults using refs like commonResponse#/definitions/notFound. However, your first use case with # at the end should be covered by this PR now 👍🏼 Once the other one's merged, it should be supported out of the box!

cc @asc11cat @Eomm @ivan-tymoshenko, tagging you folks for visibility since you were discussing on the other PR 😄

[melroy89]

However, your first use case with # at the end should be covered by this PR now 👍🏼 Once the other one's merged, it should be supported out of the box!

Let me test it again..

[melroy89]

Yup works now! Thank you so much! I was looking for this for years.

@Nelspike I would like thank you by providing a donation! Any payment method will do.

[Nelspike]

@melroy89 - Thank you for your generosity and recognition! I'm totally okay doing this without a donation whatsoever, happy to help wherever I can 🎉

[melroy89]

@melroy89 - Thank you for your generosity and recognition! I'm totally okay doing this without a donation whatsoever, happy to help wherever I can 🎉

OK, if you change your mind. Let me know.

[melroy89]

Now hopefully this PR gets merged somewhere before 2030 :)

[jean-michelet]

Now hopefully this PR gets merged somewhere before 2030 :)

I don't think this type of comment is going to help us move any faster. If @climba03003 is ok with the code, I will merge it.

[melroy89]

Now hopefully this PR gets merged somewhere before 2030 :)

I don't think this type of comment is going to help us move any faster. If @climba03003 is ok with the code, I will merge it.

No one can take a joke anymore.

[gurgunday]

lgtm

Haha don't worry @melroy89, I assure you we'll get this done this year :)

I just need another member (@climba03003 would be golden) to review

[climba03003]

Please state the $ref currently only support the format of matching $id.
It cannot lookup the fragment part. For example, schemaId#/definitions/subSchema

[Nelspike]

Good one, would pointing to the PR that's currently in the works (#676) suffice? 👍🏼

[Nelspike]

Addressed in a82f2a7, @climba03003 - Feel free to take another look!

[gurgunday]

lgtm

I can release tonight, if anyone's free, feel free to merge and release

[jean-michelet]

Release tonight plz @gurgunday!