I had two listings of Franck’s Symphony in D in compositions. One with 19 items, one with 9. As I’ve done in the past I selected both and merged, selecting the one with 19 items as the main composition. But nothing seemed to happen. After the merge I still have 2 listings of 19 and 9. Except that when I select the composition with 9 items it takes me to the list with 19 items. Selecting the composition with 19 items takes me to the same list. So those 9 albums that contain this composition are no longer there. How do I fix this? Can I unmerge somehow?
As Ludwig says, reproducibility is key. The first thing to do is find an example which can be conveniently uploaded. To date such uploads haven’t caused the bug in Brian’s system, meaning it may be associated with some state rather than the media.
A state bug can be subtle and hard to reproduce. Brian found a particularly stealthy one some time ago that was scrambling merges where a Roon Remote was restarted after the Core was running. Vova, Mike and I hadn’t been able to reliably reproduce it.
If you can identify a consistent failure to merge with particular compositions, then we’ll flag Brian and arrange an upload. If the failure is intermittent then it is more likely to be a state bug than something in the media, and the steps to the relevant state will have to be identified.
And to clarify the “bug” further: if you search for that specific “unmerged” composition, two almost identical entries are displayd under compositions in the search result. When trying to merge, nothing happens - still two slightly different entries displayed.
BUT: if you click on either of these two entries, I am shown the SAME page (so a merge seemed to have happenend on a database level!). Just for a few iterations of that compositions, the display is not … “updated”?
So correctly speaking, I think we are not talking about a merge not working but some display issue.
This is really the key @Bendix_Christians – in the case that @andybob is mentioning, the ever-so-elusive steps involved taking a series of actions, one of which involved restarting a remote at just the right moment.
That’s not to say that the missing piece here is as obscure as that – just that finding the steps that reproduce the issue reliably is generally going to be the fastest way to get this problem happening in front of a developer so it can be resolved.
Obviously, this is a tricky one if @Ludwig hasn’t pinned it down yet – the more examples you can give, and the more detailed you can be about the content involved and what you were doing when this occurred, the more likely it is that we come up with something crisp to work with.
If you can provide a few examples, @Ivan and I can do some tests and maybe we’ll figure it out. Thanks for your patience here @Bendix_Christians – these types of issues are always the most frustrating, but we usually get them eventually