What’s the turnaround time / plan for fixing MetaData errors ?
Problems like the one mentioned in this thread almost always happen because of an error in the source data from one of our metadata providers.
They get fixed when we report the error to the provider. Then the provider incorporates our requested changes, they publish them to us, we publish them to our cloud services, and eventually your Roon install pulls the updated metadata. This process can take a few weeks, especially if tracking down the root cause takes us some time.
Cases like this one are pretty tricky–an artist might be referenced in a few hundred places in each of our provider’s databases. If one of those references points to the wrong “Lamb”, it can cause the two distinct artists to be glued together in our system. Tracking down the exact point in a provider’s data where the root cause of the problem is is a pretty time-intensive/manual task.
Because it’s unsustainable for us to deal with these manually, one by one, we are working on building an automated system that identifies likely mistakes in the source data so we can submit these mistakes to the providers in bulk. This should fix this class of error broadly for the long term.
On the other hand, if the cause is a bug in our code, usually the overriding concern with metadata handling changes is the amount of risk involved–if we change the logic for how performers or albums are deemed to be equivalent to one another, the potential for collateral damage is significant. These sorts of changes need to go through a lot of testing before we can confidently release them. When we get equivalences wrong for major artists/albums, it’s very embarrassing, and there’s some risk of that sort of thing every time we tweak the logic.
In parallel with the project to search for mistakes, we are also in the process of revising how some of our metadata infrastructure works to make it simpler and quicker for us to test sweeping changes to the more risk-laden parts of the system. This will allow us to move faster in the future.
The ultra-long-term solution is to open up Roon’s database to user contribution/correction. This is a big project, and it’s not the top thing on our list right now, but I’m guessing we’ll be starting the R+D process for that sometime next year.