Steve Troughton-Smith
Those of you with subscription apps, how do you handle dropping older versions of the OS? Or am I overthinking the problem?
Feb 21 at 3:55pmIvory for Mac
Show threadBruno ScheeleFeb 21

@stroughtonsmith Without having an app of my own, you’re probably overthinking it. The people you dropped might notice after a while and be slightly miffed.

But as long as they’re happy and keep paying the subscription, they shouldn’t really notice it.

Show threadBruno ScheeleFeb 21

@stroughtonsmith If you want to achieve extra good karma, you could inform them of the app’s “Won’t be updated anymore” status and let them decide if they want to keep the advanced features by keeping the subscription.

As long as you’re not holding previous work hostage, both options are fine, imo.

Show threadMartin HöllerFeb 21
@stroughtonsmith I don’t think much about it. I support OS versions back as far as is convenient for me and makes sense.
Show threadBuckFeb 21
@stroughtonsmith Please don't drop iOS 18 or Sequoia.
Show threadSteve Troughton-SmithFeb 21
@buck iOS 26 is the minimum for any major releases of my apps going forward. There's just no way I'm maintaining two separate UIs for every app
Show threadJonathan Feb 21
@stroughtonsmith @buck I’m doing the same after my next update to Snippets Studio. It’s just too much 😩
Show threadBuckFeb 21
@stroughtonsmith Sorry to hear that. Will existing versions keep working? We aren't going to get fully rug pulled are we?
Show threadSteve Troughton-SmithFeb 21
@buck should work unchanged; no rug-pulling, even the subscription version grandfathers existing purchasers
Show threadBuckFeb 21
@stroughtonsmith Thank you! LOVE and use Broadcasts constantly 🙌
Show threadChris Wu 🪐Feb 21
@stroughtonsmith Right now supporting one prior version of iOS but that was more involved this time as you well know.
Show threadDer TeilweiseFeb 21

@stroughtonsmith Will the old version still be usable or will it break due to server side changes?

Let’s say one is on iOS 18, not willing to update. If you drop 18, does that mean the app no longer works (but the subscription continues until the user notices and cancels!) or will there just be no more updates?

I think some users will be quite angry if they loose the (working) app.

No new features until they update the OS is probably okay for most users.

Show threadSteve Troughton-SmithFeb 21
@teilweise until fate intervenes. That could be a server endpoint going offline someday, or it could be certificate expiry or some other OS security change, as happens sometimes. I was curious what everybody else was actually doing in their apps; theoreticals are fine, but being chained to the oldest OS version you've ever supported hurts development, hurts the app, hurts the rest of the users
Show threadDer TeilweiseFeb 21

@stroughtonsmith My post is an user’s point of view.

If one day the (3rd party) servers no longer support the old protocols … well, sad but that happens. After all, on Apple AppStores, the devs are doomed, fixes for older versions aren’t supported, AFAIK.

Servers you control: As long as you get paid (did not cancel the subscription from your side), well, you are getting paid to keep it running … (user’s point of view)

Show threadKevin HamiltonFeb 21
@stroughtonsmith Syncing is our biggest challenge for dropping older OS versions, which we plan on doing later this year. We've extended the data model to cover our roadmap, added extension fields to each record type (basically a dictionary stored a BLOB and a 64-bit integer used as a mask). We also a generic object record type and made sure the current version handles all of this gracefully without data loss. The local data model (Core Data) is divorced from the Cloud (CloudKit).
Show threadJosh HrachFeb 23
@stroughtonsmith My big hold up was if dropping an OS release meant a user would be in a position where one device with the newer app version reads/writes to a newer data format that the old can't understand. If that's not an issue, then dropping older OS versions isn't that big of a deal. Supporting N-1 is reasonable, and if you're able to go older with no other code changes, I say go for it.