BangleJS2 post- 2v26 firmware update:: Widget display, Pattern Launcher operation issues when BT disconnects ;; As normal when BT-connected.

[m3iow3r]

Hi y'all,

Anyone else seeing strange issues post- 2v26 Firmware (FW) update?

• I have tried Reinstall all apps to no avail.
• Stupidly, I didn't backup until after the FW update!
  • QUESTION:: Haven't tried a 2v25 FW downgrade -- I imagine this wouldn't be problematic, as it's a FW-only change. Am I right in thinking this?
• QUESTION:: Any further debugging / logging to be investigated? (I'm new to Bjs2; comfortable with code/dev).

Thanks!! =^..^=

Strange issues::

• Operates as normal when BT-connected.
• When (auto / manually) disconnected::
  • Fails to reconnect.
  • Appears / works normal until an app is opened (e.g. Settings ;; Pomodoro ;; SimpleTimer -- seemingly any app though) then::
    • Widget layout "corrupts".
    • Pattern Launcher stops responding from clock.
    • Potentially interfered with Run+ / Recorder GPS logging..?
    • Maybe some other things!
    • Can still access apps via Launcher.
  • Full reset returns to BT connect-able. As does changing Debug setting with subsequent auto-reboot.
    • Issue re-occurs with subsequent BT disconnect.
  • Issue occurs when no BT available (i.e. BT turned off on Phone, Laptop)).
  • Exception logged on "soft reset" (which doesn't resolve issue)::

Uncaught undefined

• Log procedure:: [[ BT connected ;; Debug enable ;; BT disconnect ;; issue manifests ;; Reset device ;; Debug disable ;; Extract log ]]
  • Log captured is unenlightening;

    <- Terminal
    �[?7l
     ____                 _ 
    |  __|___ ___ ___ _ _|_|___ ___ 
    |  __|_ -| . |  _| | | |   | . |
    |____|___|  _|_| |___|_|_|_|___|
             |_| espruino.com
     2v26 (c) 2025 G.Williams
    
    Uncaught undefined
    > �[JUncaught undefined

• Also logged::
  • However it seldom appears when the behaviour above is experienced.

    > �[JUncaught Error: ERR 0x9 (INVALID_LENGTH) (:2840)
     at line 33 col 157 in .boot0
    ...ising(a(Bangle.bleAdvert),b)
                                  ^
    in function "advertise" called from line 33 col 557 in .boot0
    ...,a.bleAdvert=i);advertise(l)
                                  ^
    in function "set" called from line 34 col 37 in .boot0
    ble_advert).set(6159,[E.getBattery()]);
                                        ^
    in function called from system
    > �[JUncaught Error: ERR 0x9 (INVALID_LENGTH) (:2840)
     at line 33 col 157 in .boot0
    ...ising(a(Bangle.bleAdvert),b)
                                  ^
    in function "advertise" called from line 33 col 557 in .boot0
    ...,a.bleAdvert=i);advertise(l)
                                  ^
    in function "set" called from line 34 col 37 in .boot0
    ble_advert).set(6159,[E.getBattery()]);
                                        ^
    in function called from system
    >Button held down - interrupting JS execution...
    �[JExecution Interrupted
    > �[JUncaught undefined
    >Button held down - interrupting JS execution...
    �[JExecution Interrupted
    > �[JUncaught undefined

Images

(( Camera pics 'cos unable to g.dump() when failing to BT connect ))

• Normal UI::
  
• "Corrupted" Widgets::
  • NOTE:: HRM widget now aligned to left
    
• "Corrupted" Widgets inside Pomo app::
  • NOTE:: LIGHTSWITCH widget now aligned farthest-right (whereas CLOSE is set to farthest-right) ;; HRM widget now aligned to left.
  • The redraw issues I think are just due to Debug printing -- they don't appear as part of the issue.
    
• Pattern Launcher unresponsiveness unable to be imaged; BT connect fail likewise! ;-)

[gfwilliams]

Thanks for all the debug info. That really helps. Do you have BLE GATT Battery Service installed? https://banglejs.com/apps/?id=bootgattbat

I think that app is trying to advertise your battery percentage over Bluetooth (do you actually need that?) but 2v26 advertises some extra data (intended to make the Bangle easier to find/pair) and it's probably meaning that when the extra data is added the advertisement gets too big and it now fails.

I think that may be the only issue - that the code runs and fails, and then subsequent code doesn't run properly after that. If you uninstall BLE GATT Battery Service it should work fine I think?

Haven't tried a 2v25 FW downgrade -- I imagine this wouldn't be problematic,

No, downgrading should be fine. Also the upgrade doesn't touch what's in your Bangle's external flash memory so really the code there should be as before even if you did the backup after.

[m3iow3r]

Just FYI I think you must have some other things advertising data too? It seems to work ok if it's just BLE GATT Battery Service

Don't think so? Can't spot anything obvious via App Loader / app list ( https://github.com/orgs/espruino/discussions/7764#discussioncomment-13204774 )...

I've added some code that disables the manufacturer advertisement that 2v26 added if you add advertisements (it also shouldn't crash in future, just show an error on the console). Maybe try re-installing all your bluetooth advertising apps from the development app loader at https://espruino.github.io/BangleApps/ and see if that helps?

Thanks -- will give a go and report back at next opportunity 👍

[m3iow3r]

Just FYI I think you must have some other things advertising data too? It seems to work ok if it's just BLE GATT Battery Service

Don't think so? Can't spot anything obvious via App Loader / app list ( https://github.com/orgs/espruino/discussions/7764#discussioncomment-13204774 )...

Still not sure what this could be, sorry! I haven't looked at inspecting the GATT traffic or anything, so will update if I ever do.

I've added some code that disables the manufacturer advertisement that 2v26 added if you add advertisements (it also shouldn't crash in future, just show an error on the console). Maybe try re-installing all your bluetooth advertising apps from the development app loader at https://espruino.github.io/BangleApps/ and see if that helps?

Confirmed no crash with the Dev App Loader version of bootgattbat. Nothing enlightening beyond the Espruino header dumped to the logs (when BT disconnected and opening apps -- the same reproduction protocol); see below.

I'm just using the debugging method (Settings => Utils => Debug => Both) from Debugging Tips - Espruino .
QUESTION:: Are there additional steps required to get meaningful output in the logs? Debug version of the App Loader or something? -- a quick websearch yielded nothing promising...

Thanks! (and, closed from my perspective.) =^..^=

    >�[?7l
     ____                 _ 
    |  __|___ ___ ___ _ _|_|___ ___ 
    |  __|_ -| . |  _| | | |   | . |
    |____|___|  _|_| |___|_|_|_|___|
             |_| espruino.com
     2v26 (c) 2025 G.Williams
    
    >�[?7l
     ____                 _ 
    |  __|___ ___ ___ _ _|_|___ ___ 
    |  __|_ -| . |  _| | | |   | . |
    |____|___|  _|_| |___|_|_|_|___|
             |_| espruino.com
     2v26 (c) 2025 G.Williams
    
    > �[J-> Bluetooth
    <- Terminal
    �[?7l

[gfwilliams]

Thanks! If you get a moment it'd be interesting if you typed Bangle.bleAdvert in the console so we can see what's in there.

QUESTION:: Are there additional steps required to get meaningful output in the logs?

No, I'm afraid not. Generally apps don't have logging built in - when an error is thrown and not caught it's printed to the console, but we don't have anything like switchable loglevels - although that might be a neat addition!

[m3iow3r]

Thanks! If you get a moment it'd be interesting if you typed Bangle.bleAdvert in the console so we can see what's in there.

No probs -- any additional steps let me know! =^..^=

No GATT installed::

Bangle.bleAdvert
=undefined
> 

Official App Loader bootgattbat installed (and exhibiting above issue)::

>Bangle.bleAdvert
={
  6159: [ 70 ]
 }
> 

Dev App Loader bootgattbat installed::

>Bangle.bleAdvert
={
  6159: [ 64 ]
 }
> 

QUESTION:: Are there additional steps required to get meaningful output in the logs?
No, I'm afraid not. Generally apps don't have logging built in - when an error is thrown and not caught it's printed to the console,

OK thanks. Makes sense, especially for embedded. I'm no stranger to instrumenting code!

but we don't have anything like switchable loglevels - although that might be a neat addition!

I've not looked at JS/TS implementations of this -- surely better than C/C++ #ifdef hell..! 😵‍💫

[gfwilliams]

Thanks! That's an odd one - I wonder why the official one failed and dev one didn't then...

I'm not sure JS has anything built into the spec for loglevels (apart from console.log/info/warn/error that most people implement) - I guess I could add a 'loglevel' to the interpreter that ignored .info/etc depending on the setting. The code for it would still be stored/executed though which isn't ideal.