A lot of the cloud services that you can think of, however, are going to function fundamentally different that AppSheet. For example, AppSheet (a cloud service) pushes updates that we see in the editor without going offline. But that is possible because none of the data for the editor is store on our computers. Same thing for when apps like Facebook, Google, and Twitter suddenly change their interfaces despite no app update.
You can’t really think of your app running on a cloud service as, itself, a cloud service and therefore innately having cloud service abilities. Part of the reason that makes the cloud, the cloud, is because data is multiplied and stored in several different places (kind of like a RAID setup, but different). Because of that when servers do down for maintenance or glitches, a lot of time you won’t even notice. It also allows them to do upgrades without shutting down, because they can roll the upgrade out slowly with two concurrent versions running side-by-side until everyone is switched. Not to mention they can also run custom scripts that move data automatically for people/users in the event of a structure change.
Your app and your data doesn’t exist in that same way that the underlying platform does. Part of the problem is that, how is AppSheet supposed to understand that it’s still ok to sync the data even though the column structure has changed? In this example, it’s only new columns, and therefore the old data still fits into those columns. However, AppSheet has no way to verify that this is true, since AppSheet is just a middleman, so to speak. The middleman can’t interpret your intents when you change the backend its working with.
I believe giving some sort of “toggle” for this behavior would introduce worse problems, when in reality you would need much more granular control for how AppSheet deals with errors and sync issues. You basically want some form of TRY/CATCH system, which I highly doubt AppSheet can give us, at least any time soon. And beyond that, it starts to get really far away from the no-code idea.