[JakePre]

What and why

The iOS side of the Switch 2 controllers (#1267), as you asked: the iOS front end starts the shared transport, pad_switch2_apple.m, as front_mac.m does.

• DmGamepad.m gives a Switch 2 pad a player slot the way front_mac.m does:
  • the same slots as the GameController pads;
  • its rumble goes to dm_switch2_rumble. With DM_PAD_MERGE, player 1's rumble follows the controller last used, as on the Mac since your landing;
  • press to join (DM_PAD_PRESS_TO_JOIN);
  • the controller applet's prompt: a pad's A and B count, a pad that hasn't joined counts as waiting, and no input reaches the game while the prompt is up;
  • the touch controls hide while a pad is in use, and the transport pauses while the app is in the background.
• The iOS Info.plist gets NSBluetoothAlwaysUsageDescription (project.yml). Without it, the transport stays off.
• settings.tsv offers DM_SW2 on iOS too, and test_launcher_settings.py no longer lists it as Mac-only.
• docs/playing.md and docs/settings.md say how it works on iOS.

iOS doesn't bond. iOS gives an app no way to read the device's own Bluetooth address, which bonding writes to a controller. Your IOBluetoothHostController read is Mac-only. So a controller in Sync is used without bonding, as pad_switch2_apple.h already says: hold Sync after each start of the game. A button press alone doesn't wake it toward the phone. The player docs say this.

The commit carries Front-ends: new:. An iOS app that offers DM_SW2 needs the Bluetooth usage description in its Info.plist.

Checked

Its host part, #1907 (Switch 2 controllers with press to join), landed as 7ef8339, and this branch is rebased onto main with it: one commit, the iOS wiring. The iOS front end's s2pad_attach already returns whether it keeps the pad, as pad_switch2_apple.h now asks.

• Tests:
  • test_launcher_settings.py, test_settings.py, test_docs.py and dm/ios/tests pass.
  • DmGamepad.m compiles for iOS (arm64, -mcpu=apple-m1) with no warnings.
• iPhone 17, 8 November, this branch: Super Mario Galaxy (010099C022B96000, v327680), a left and a right Joy-Con 2, with Switch 2 controllers and Press a button to join on.
  • Sync on each: both connected and stayed connected (without Switch 2 controllers with press to join: a controller that connects waits for its first button press instead of being let go #1907's fix, each was dropped at once).
  • The first press: a button pressed: player 1, the two as one pair.
  • SL + SR split the left one off as player 2, and L + R made them one pair again.
  • No bonding, as expected: this host's Bluetooth address is not known: … needs Sync again next time.
• iPhone 16, 7 November, the earlier wiring: Paper Mario: The Thousand-Year Door (0100ECD018EBE000, v65536) and Super Mario Galaxy, docked with the gyro pointer. Pairing by Sync, motion and rumble worked, and the touch controls hid while the Joy-Cons were connected.
• Hold Home to quit: not on iOS, as for every other controller there. The app gives Home and Capture no function and only keeps iOS from taking them as system gestures.
• Not checked:
  • an iPad;
  • the Pro Controller 2 on iOS;
  • DM_PAD_MERGE with a Switch 2 pad on iOS;
  • the controller prompt (Galaxy doesn't open it).
• regress.py: not run. This PR's own change is the iOS front end, its Info.plist, one settings row and two docs.

Docs

• docs/playing.md (a section for iPhone and iPad) and docs/settings.md (the setting moves from the Mac window's list to Controls, and says what differs on iOS); no symptom entry, since nothing broke
• No chain tool changed
• No game files, keys or data extracted from a game

🤖 Generated with Claude Code