💡⚡️🍾 Espruino 2v27 Released

[gfwilliams]

I didn't do a release note for 2v26 (which was released only a few weeks ago) so I'll try and do a bit of a round-up here of what's been added across both versions. It's quite a big one!

Debugging

• Adding support for console.trace, (new Error()).stack and much better backtraces in general, which now use file:line:col format
• Out of memory errors can now print a backtrace of the exact location the error happened
• Debugger now works well with files executed entirely from flash
• E.setFlags({onErrorFlash:true}) flashes the red LED for 200ms on any uncaught errors
• E.setFlags({onErrorSave:true}) saves any uncaught errors to a file in Storage called ERROR
• If debugging via bluetooth and we disconnect, quit debugging (locks Bluetooth otherwise)

Espruino General

• 8/16 bit ints can now be pretokenised for faster execution
• Templated strings without Templates now treated as normal strings (faster)
• ioBuffer FIFO changed from 4 byte blocks to variable length blocks. Far more efficient for USB/BLE/other block data
• Unlock functionCode early in function execution to reduce locks needed during recursion
• E.defrag() now defragments everything, including Flat Strings
• FAT filesystems: Added r+ file open mode (and allow w+ to read), if we can't allocate file, attempt defragmentation and try again
• Graphics.asImage() can now take x/y/w/h options to make only part of the graphics into an image
• Increase chars shown in strings in console to 60
• nRF52: Fix unreliable internal flash writes when writing >2k blocks (and also if Ctrl-C issued during a write)
• nRF52: If we reboot because of LOCKUP, don't enable DCDC converter
• nRF52: Improve software timer accuracy by keeping timer running while executing jstUtilTimerInterruptHandler
• Jolt.js/Bangle.js 2: Move builds from 6.0.0 to binary-compatible 6.1.1 softdevice
• Puck.js: Fix Puck.magTemp() for hardware 2.1
• Puck.js: On v2.1 ensure Puck.mag behaves like other variants - just returning the last reading (avoids glitches when used with Puck.magOn)
• Storage: Don't attempt to compact if it's obvious that we don't have space for the file even when compacted
• Storage: Ensure compaction removes any files with invalid file types

Bluetooth

• All Bluetooth Espruino devices now advertise the Espruino 0x0590 manufacturer ID by default (more reliable than scan response)
• Bangle.js/Jolt.js: enable BLE Long Range connections
• Bangle.js/Jolt.js: Allow 1mbps,coded/coded,1mbps and up to double-length advertising in 'extended' advertising mode
• Bangle.js/Jolt.js: Allow scan parameters to be passed to connect (enabling coded phy connection)
• Increase Packet timeout from 1 to 5 sec (increase reliability over slow Bluetooth LE connections)
• Ensure Bluetooth notifications work correctly when two separate connections use the same handle for their characteristics
• nRF5x: Fix BLE connection timeout (was 40ms, not 4s) for more reliable outgoing connections

Bangle.js

• Added new Renaissance fonts and g.findFont function for selecting the best font (this helps a huge amount with the readability of menus)
• E.showPrompt has bigger buttons and better alignment
• 'health' data now included bpmMin/bpmMax rather that just BPM
• Fix Bangle.js 'internal flash' builds

Other boards

• ESP32: Remove FlashFS, adjust partitions to increase available Storage from 256kB to 896kB
• Linux SDL build: mouse events now create E.on("touch" events like on CYD
• nRF52840: Add E.getVDDH() method to get VDDH voltage
• nRF5x: Allow entering UF2 bootloader mode by calling E.rebootToDFU() (on boards that have such a bootloader)
• Added board: Seeed XIAO BLE

... plus a whole bunch more! Check out https://www.espruino.com/ChangeLog for more information as well as numbers of GitHub bug reports related to the changes.

[hcarrega]

Since this last update I have had immense problems I have not had since I bought the watch.
After the 2v27 installation my watch in 1 day was out of battery, when I woke up I was turned off and I got the load and it came back to life, but it's almost always spent a while to be blocked and I have to carry on the button to make rebootja I did two factory resets to see if I solved something but nothing, which I can do
thanks
Usually I don't have the watch always on the phone call once a day or 2 days to do sync

[gfwilliams]

I'm sorry you're having issues, but 2v27 has has quite a bit of testing, and I've been running it here for some time without issues.

When you did the factory reset, did you reinstall/restore anything or change settings? Or you're saying that with a completely standard, factory reset software you now have battery life issues where before it was fine?

You could try reinstalling an older firmware and see if that fixes it for you, just to rule that out?

Also if you connect with the Web IDE at https://www.espruino.com/ide and use the watch, do you see any error messages printed?

Could the change in battery life correspond with something else that happened, not just the firmware update?

[hcarrega]

I did a full reset and only stays it default apps , didn't use any backup .today during the night the watch get stuck again , I disable heat rate to see if is the problem if it still the same I try to revert to 2v25.
On 2v25 I get more than 2 weeks . Gonna test more and I will ping again if I found something

[gfwilliams]

Ok - by default (if you just did a factory reset) heart rate should be disabled I believe.

But yes, please try 2v25. Maybe also connect with the IDE and run E.getPowerUsage() in the left-hand side and post up the result?

[perinoid]

Since this last update I have had immense problems I have not had since I bought the watch. After the 2v27 installation my watch in 1 day was out of battery, when I woke up I was turned off and I got the load and it came back to life, but it's almost always spent a while to be blocked and I have to carry on the button to make rebootja I did two factory resets to see if I solved something but nothing, which I can do thanks Usually I don't have the watch always on the phone call once a day or 2 days to do sync

Same for me. Additionally, the watch seems to hang often - no watchface updates, no way to enter the menu, nothing. It was enough to for instance select activity synchronization, but also when some of the messages were to be delivered. I guess, it was the moment, when the massive battery drained happened. Only the general reboot helped but only for a moment.
I tried to downgrade to 2.25 but it wouldn't even boot after. (errors like "Uncaught SyntaxError: Got ?[212] expected EOF".
Right now I'm on 2.26 but it doesn't seem to be any better. It behaves exactly like 2.27 so my guess is, the problem appeared somewhere between 2.25 and 2.26 and remains in 2.27.

[gfwilliams]

Hi - Please can you go to the App Loader, click More... and then Reinstall apps - that should rewrite the app's code without changing any settings or deleting any stored data. That should make it work fine in 2v25.

But: I think you might have been unlucky when you updated to 2v27 and there was a bug in the App Loader where some apps it uploaded could get slightly corrupted. So if you go to 2v27 and do the Reinstall apps again you'll probably find it works perfectly fine now

[hcarrega]

i get a new hangup, now revert to 2v25

[ihewitt]

Great, thanks for the update! I'm still struggling to see what could have caused the increased power draw.

Please can others let me know if you have this issue? Specifically if you're in the UK I could swap your Bangle out for a new one and then I can take it to bits, check the power draw, and try different firmwares until I find what's causing the increased power consumption

Argh, didn't see this request until I fixed mine! So, I saw this power draw twice, running flat overnight.. the first time I assumed I'd left something running, but then it happened again... So I manually uninstalled lots of stuff (I had a lot installed) and it's been fine for several days now with negligible power drop. But obviously I wasn't paying attention to exactly what I removed. I can try to reinstall what I think was on and see if I can get it to happen again.

[hcarrega]

In what version you are?

[ihewitt]

went from 25 to 27 and had the power drain.

[hcarrega]

Same here. Now back on 25 and works normal

[gfwilliams]

@ihewitt thanks for the update! I don't suppose you had a backup of the old install that had the power drain do you?

If you have one, or remember/find the app that caused the increased power draw I'm very happy to look into it. It's just a shame I'm not seeing it here.

To others: if you're experiencing this power draw please can you make a backup of your device and send the backup to me? Hopefully by putting that backup on a device here I can reproduce and track down the power usage.

[perinoid]

Can someone please help me downgrade to 2.25? 2.26 and 2.27 seem to be unusable on my Bangle2 (all the same complains as above) and when I'm trying to downgrade, after the firmware is installed I'm getting only something like this (both on the screen and in the console, which - fortunately - still works).

[perinoid]

I just replied to the other post you made at https://github.com/orgs/espruino/discussions/7781#discussioncomment-13662509 with the answer...

I'm pretty sure you just need to click Reinstall apps in the App Loader. It'll make 2v25 which you're on work, but I think likely if you re-install 2v27 then it'll fix that too

I'm trying - but as soon as I select "Reinstall apps" and confirm, Bangle goes to a kind of "about" screen and stops responding. The app shows a progress bar, which doesn't change... forever. I'm trying with 2.25 now (downgrade first). We will see.

EDIT
With 2.25 I also cannot reinstall apps. Bangle goes to "Updating boot0..." and stays there, while the app only shows a persistent, not changing progress bar.

One more EDIT
I removed all the apps from the watch, downgraded, then reinstalled the apps and now it seems to work fine under 2.25. All the options that made 2.27 crash earlier, now work fine.

[thyttan]

I've also seen my bangle lock up and reboot a couple of times recently, say a week or two. Mainly when it's been sitting without interaction and I am about to use it again. Sorry I haven't looked into it more than that for now..

My main suspect was the new clock face I'm using (Modern Clock). But maybe it's something else with the report here.

At some times when I intend to use the watch no input from either hardware button or swipes (quicklaunch) works to navigate away from the clock face. But the system is not necessarily locked up because I can still execute commands in via the Web IDE.

So that's a bit different.

My next steps is to reinstall the apps again and see if that helps. Then maybe downgrading fw to see if problems goes away that way.

[gfwilliams]

@thyttan Maybe try reloading the clock (with load()) with the IDE connected and see if any errors are reported? It sounds like maybe some JS software is causing an error and meaning that the button/touch handlers aren't being registered properly.

@perinoid glad it's working again - I guess it's possible the storage got corrupted somehow (or completely full?) and that caused problems with the Updating boot0... screen.

[thyttan]

Maybe try reloading the clock (with load()) with the IDE connected and see if any errors are reported? It sounds like maybe some JS software is causing an error and meaning that the button/touch handlers aren't being registered properly.

No errors reported when investigating now when it happened again.

Description of console in-/output below:

1. I noticed the problem was present. No quicklaunch swipes or HW Button working to switch apps.
2. Bangle["#onswipe"] to see reported registered handlers.
3. load() to reload the clock face (modern clock).
4. Bangle["#onswipe"] again to see if the list is same as on step 2 (seems like it).

Bangle["#onswipe"]
=[
  function (lr,ud) { ... },
  function (lr,ud) { ... },
  function (lr,ud) { ... },
  function (_,numOrObj) { ... },
  function (lr,_) { ... },
  function (force) { ... },
  function (lr,ud) { ... }
 ]
>loaf()
Uncaught ReferenceError: "loaf" is not defined
    at REPL (:1:1)
loaf()
^
>load()
=undefined
>Bangle["#onswipe"]
=[
  function (lr,ud) { ... },
  function (lr,ud) { ... },
  function (lr,ud) { ... },
  function (_,numOrObj) { ... },
  function (lr,_) { ... },
  function (force) { ... },
  function (lr,ud) { ... }
 ]
> 

I have a hunch this may be happening mostly after midnight. Seems like a weird factor for this bug but maybe it's relevant somehow.

Installed
sched (0.33), kbmulti (0.09), reply (0.03), widmessages (0.07), messages (0.67), widmsggrid (0.07), msgwakefup (0.02), delaylock (0.01), notify (0.14), widminbate (0.05), podadrem (0.11), android (0.44), widanclk (0.03), backswipe (0.05), torch (0.11), calculator (0.08), widbt_notify (0.18), smpltmr (0.12), clkinfostopw (0.03), runplus (0.30), dtlaunch (0.28), quicklaunch (0.16), kineticscroll (0.05), widalarm (0.02), alarm (0.52), edgeclk (0.05), autoreset (0.02), chargent (0.07), fastload (0.06), ateatimer (0.02), drained (0.08), recorder (0.47), boot (0.66), sleeplogalarm (0.06), msgtwscr (0.01), messageicons (0.09), gpsinfo (0.13), tileclk (0.01), worldclkinfo (0.01), setting (0.80), jwalk (0.01.1), spotrem (0.14), clock_info (0.18), health (0.35), sleeplog (0.20), modclock (0.04), forge (0.03), messagegui (0.88)

[gfwilliams]

There is an on('midnight' event and you could check for handlers there in case an app is screwing something up?

Might also be worth doing:

require("Storage").write("saveerror.boot.js","E.setFlags({onErrorSave:1});");

This will enable writing the first error found to flash - when you next connect with the IDE and reset/load it'll tell you there's an error if there is one and how to view it. It might help to track down if a JS error is causing your problems

[enaon]

the debugging features are very nice, well done Gordon.