Clarification on result response from scrobble

[cquike]

The scrobble endpoint (https://opensubsonic.netlify.app/docs/endpoints/scrobble/ ) returns an empty subsonic-response "on success".
The specification does not explains what success really means. If the "id" of the media file is not found, is it still a success?
I had the problem, using the ampache server, that I scrobbled with curl a song with a wrong id that is not present in the database. ampache was still giving an answer with status "ok". Is that expected from the specification point of view? If so, does this means that there is no way for the client to know whether the scrobbling was successful or not?

To complicate the things, if the client provides several id's there is no way with an empty response to know which ones failed and which ones were successful.

[Tolriq]

Server should return "70 The requested data was not found."

[cquike]

Thanks for your answer!

In the case of several id's in the request, should the server return that when any of the id's is not present?

Also, just to clarify how that return is done: Is that encoded in a subsonic-response with status "failed" and error (https://opensubsonic.netlify.app/docs/responses/error/) set to 70?

[paulijar]

Yes, that's how errors are returned in the (Open)Subsonic API.

It would actually be good idea to document, what should happen if one of multiple given IDs is bad. Like is the response "ok" or "failure" and do the good IDs gets scrobbled or not. I'm afraid that there may currently be differences between distinct servers. For my own server (Nextcloud Music), I can tell that currently it scrobbles the given songs one by one, and stops and returns "failed" if hitting an invalid ID. That's actually quite bad way to do it, and I would like to improve it.

[cquike]

Thanks @paulijar for your answer!
Yes, I agree that it would be a great to document the different use cases in the subsonic specification to avoid ambiguities between client and servers.
For ampache I have created a PR that changes the behavior and fails if any of the id's is invalid. In that case nothing is changed, not even for the valid id's: ampache/ampache#4328

[paulijar]

fails if any of the id's is invalid. In that case nothing is changed, not even for the valid id's

Yes, this is probably the most sane way to handle it.

[cquike]

I have created a PR for the clarification #202 . Suggestions welcome!