Something went wrong while fetching comments. Please try again.
Created by: courier-new
Erik's comment on #23966 got me thinking... since we can't take any action on already-published changesets anyway (we can't convert them to draft nor unpublish them), we might as well just exclude them from what's selectable (and then explain what's up in that FAQ entry).
Of course, we actually have a lot of potential changeset states. Do correct me if this feels like it's going too far, but it seemed to me like maybe we shouldn't allow modifying the publication state of a changeset unless it's unpublished or a draft: every other state is either some form of "already published" (open, closed, merged, deleted) or "potentially about to be published" (failed, retrying, scheduled, processing).