Same as with STYLE-74, which I think is a subtask of this. (I don't know what keeps that RFC, or what you call it nowadays, from passing.)

This is a serious regression; just consider what was lost during NGS migration and the thousands of merge edits in the aftermath where medium-specific information was simply swallowed.

We might also need a new report when this has passed: Let's say two mediums are merged and medium 2 is appended to medium 1, each disc has a release AR saying that some orchestra performed on it. The resulting release must not have a release AR where the new medium attribute, created during the merge process, is dropped, but in a release AR with medium attributes for mediums 1 and 2 because the release is not necessarily complete. (Some editor might add a third disc to which that AR doesn't apply.) The report would list those releases that have as many medium attributes as mediums. Editors must then check whether the release is complete and remove the attribute manually.

Edit: See for instance http://musicbrainz.org/edit/31629475 What am I supposed to do there? Monteverdi Choir did not perform on all 30 mediums, but I don't know either whether they performed on each and every track of the medium it once applied before I merged the discs, so that I could add it at the recording level. It is not my job as someone who regularly merges releases to sort that out; instead, that type of information added by the original editor must still be available post-merge without my having to tinker with it.