You can test this by enabling every type of notification for every entry in kcm_notifications, and then shutting-down:
Even if an applications instructs Plasma to display a notification with no content, is there ever any point to permit this?
Possibly because someone else might argue to the contrary, that the job of the shell/notification service isn’t to make judgments on the content of notifications that applications choose to send, as long as those applications had been granted the permission to send notifications?
(I don’t know enough about how notifications really work to understand how that happens, but you asked about whether something should be “permitted” on a FOSS desktop, and it seems there’ll always be at least a few folks out there insisting that the thing you’re talking about is a critical part of their workflow - perhaps the empty notification is someone’s spacebar heater 
)
Edit to add: I do generally agree with limiting/not permitting things that could only cause objective negatives for the user (like, don’t let someone break their hardware with a setting) - in this case, I’d just wonder if there’s harm occurring that would be prevented by blocking these, if that makes sense?
2 Likes
@johnandmegh, now that you mention that, I can see it being useful when testing notification support if you’ve not read the documentation stating that it must have content, I suppose. That still seems somewhat niche.