Search results for:

Hello folks!

We may have a memory leak with Module.removeAllCached(), which is pretty ironic since I hoped to use it and free some memory ;P

A bit of context. I'm trying various means to overtake the deadly 10000 character limit.
From the last month experiments, this is roughly the maximum number of symbols your application could have to fit in the Bangle's RAM.

I found a nice way to split my code into several chunks, each being a file.
My ES6 source code is leveraging dynamic imports, which instructs Rollup to produce several chunks.
A small Rollup plugin of mine turns the chunk into Bangle's files.
The app is written so screens are stored as files. Displaying a screen is simply requiring the relevant file, and calling the build() function it exposes.

Here is the code,
and here is the minified, split output that you can run on your watch.

Now if you launch the app, and cycle through the various screens with middle button, at some point you should have a LOW_MEMORY error.

Since object returned by require() are cached, I've tried to call Module.removeAllCached() after a screen was required, so it could claim the space from the previous screen.

Here is the minified code, same as previously but with the call.

You'll be surprised to get the LOW_MEMORY error the very first time you load next screen.
Few more experiments makes me believe that Module.getCached() also have similar side effect.

Could it be that listing cached modules (which removeAllCached() starts with) is eating more RAM that it should?

I lost quite some time trying to replace octal sequence with regex... with no luck, hence why I used encodeURIComponent instead.

I'll be very interested to learn about the solution, would some regex wizard lit my lamp.

You where right @Gordon.
After I've updated the boot loader, the numbers are closer.

Running the code:

> initial 694
1053
1053
1053
903
905
916

Running as an App

> initial 771
974
974
824
856
856

To give a bit of context: I would like the ability to list files present on the board.
I though it would be a small addition ton BangleApps to display such list, and allow to download them.

I've mitigated the encoding issue:

      // this is on the BangleApp:
      Puck.eval('require("Storage").list().map(encodeURIComponent)', (files,err) => {
        if (files===null) return reject(err || "");
        files = files.map(decodeURIComponent)
        console.log("listFiles", files);
        resolve(files)
      });

Thanks for the clarification folks.

This final character makes Puck.eval() failing when transfering these strings:

Puck.eval('require("Storage").list()')

will fail on:

Unable to decode "[\"+mclock\",\"-mclock\",\"*mclock\",\"+setting\",\"-setting\",\"=setting\",\"*setting\",\"+astroid\",\"-astroid\",\"*astroid\",\"+gpstime\",\"-gpstime\",\"*gpstime\",\"+compass\",\"-compass\",\"*compass\",\"+sbt\",\"=sbt\",\"+sbat\",\"=sbat\",\"+files\",\"-files\",\"*files\",\"+hrm\",\"-hrm\",\"*hrm\",\"+gpsinfo\",\"-gpsinfo\",\"*gpsinfo\",\".trishig\",\".bootcde\",\"+speedo\",\"-speedo\",\"*speedo\",\"@setting\",\"@activi\\1\",\"-activit\",\"+activit\",\"test\\1\"]\r", got SyntaxError: Unexpected number in JSON at position 357

Having chosen an octal escape sequence make it hard to (de)serialize.

Hi there!

Before opening a bug on the tracker, I wanted to double check this behaviour.
This code opens a file in append mode, and writes some text in it.

const Storage = require('Storage')
const name = 'test'
Storage.erase(name)
const file = Storage.open(name, 'a')
console.log('-> files', Storage.list())
// file not in the list
file.write('whatever')
console.log('-> files', Storage.list())
// file exists, but has a \1 octal escape sequence appended to its name

this gives:

 ____                 _
|  __|___ ___ ___ _ _|_|___ ___
|  __|_ -| . |  _| | | |   | . |
|____|___|  _|_| |___|_|_|_|___|
         |_| espruino.com
 2v04 (c) 2019 G.Williams
>-> files [
  "+mclock",
  "-mclock",
  "*mclock",
  "+setting",
  "-setting",
  "=setting",
  "*setting",
  "+sbt",
  "=sbt",
  "+sbat",
  "=sbat",
  "+files",
  "-files",
  "*files",
  ".trishig",
  ".bootcde",
  "@setting",
 ]
-> files [
  "+mclock",
  "-mclock",
  "*mclock",
  "+setting",
  "-setting",
  "=setting",
  "*setting",
  "+sbt",
  "=sbt",
  "+sbat",
  "=sbat",
  "+files",
  "-files",
  "*files",
  ".trishig",
  ".bootcde",
  "@setting",
  "test\1"
 ]

This \1 at the end: is it expected?

Thanks Conor.
I'll have a try!

Hi Robin.

I've update the links above:
https://www.espruino.com/ide/?gist=9df12­9529bccc133f9f82bf7b39de48f&upload
will write the minified app (8485 characters) straight to your bangle and run it.
The very first instruction displays number of free blocks.

Here is an example of a run (press button 2 to start, button 1 to pause, then button 3 to stop and reset):

>> initial 694
1053
1053
1053
1053
1053
1050
1050
1050
1050
1050
1050
902
932

Now, the same code, but saved in Storage under "activities" name:
https://www.espruino.com/ide/?gist=3fddc­28e3188fd5f5cded6aa1d037cb0&upload

Long press Btn 3 to get the clock, then goes to settings, select "activities" and start/pause/stop

> initial 496
699
699
699
699
699
696
696
696
696
696
696
696
696
546

I've manage it with a better minification (thank you "terser's mangle" option!).

One thing I'd like to confirm.
With IDE minification disable, when I load my code straight to the bangle, process.memory().free returns 1023 when the app starts.
When I store the exact same code on the bangle storage, and start the app with the launcher, the number of free blocks drops to 692.

Is the difference due to the app launcher itself?
Is there some memory I could reclaim?

So I'm a runner. And as I turn older, I started to run in hills and mountains.
2 years ago I spent 300€ for a Suunto smart watch, to guide me on tracks:

I really like it, but it could be improved in a lot of ways, in particular its software. At some point, it was possible to write custom screens with some sort of JavaScript :grin:
But the whole Suunto ecosystem is very closed, and their support is, for the least... not responsive.

So when I learned about Bangle.js you can imagine how excited I was.
The plan is to mimic Suunto's sport mode:

• start an activity
  
• cycle through several screens, showing metrics (current & average speed, distance, BPM, burnt calories, incline/decline, altitude, for the total activity or current lap)
  
• pause and stop
  
  

In my craziest dream, I'd love to write a navigation screen to guide you along some GPX track. But this may be too ambitious given the hardware.

Not surprisingly, writing such app is not in indoor activity. It depends too much on GPS to stay at home.
So yesterday, it was the première for my app: after work I took my bike (10km +200m) to meet my fellow runners and we went for a run across hills (11km +400m), a watch on each wrist.
It was night, raining and wind blowing.

The Bangle behave very well!
The LCD was lit up all the way long (3h30), 10 FPS refresh rate, GPS and HR on, and it ate 60/70% of its battery (same as the Suunto!).

GPS fix took a while compared to Suunto, but I think the accuracy is better (I'll perform more investigations). The lack of accuracy is my greatest complain. GPS Altitude (benchmarked against signs mentioning it) is very accurate, I suspect instant speed to be quite good as well.

No surprise: HRM gives terrible results (either 50 or 200 BPM). My skin is quite clear, and my veins very easy to spot. I was careful not wearing it on bones, and as it was night, the amount of light pollution was limited.

Despite this little issue I was amazed how good Bangle did, especially when you consider the hardware is 5 times cheaper! Amazon work @Gordon!

I'll continue crafting this sport mode, and share the code when it will be ready, would someone be interested to use Bangle for sport!

Hey folks!
Thanks for the quick answers.

few takeaways:

1. splitting my app in chunks when writing it in storage is not a good idea: if it can not be saved in the first place, it means it won't fit in RAM when read

2. as I have my own pipeline to bundle and minify my app, I will disable IDE minification in the future. This way it will be easier to get consistent results between stored app & REPL app.

3. I've reached my limit as a web developer: the way I approached my app, with beautiful and reusable classes, icons as images, and intensive event usage, does not work in such constrained environment. I'll be more pragmatic now

@allObjects my source code is split in different modules, and I use rollup & babel to transpile and bundle them in one file. Is there a way to split the app on the Bangle itself, using different "files" in storage?

Hello there!

When I upload my code from the Web IDE to the watch, it runs as expected.
https://www.espruino.com/ide/?gist=9df129529bccc133f9f82bf7b39de48f&upload
(shows you current time and start message, until you press button 2 which stops everything)

I am trying now to store it as an app.
The code string is big (18 156 chars), so it is split in chunks and writes gracefully in file.
https://www.espruino.com/ide/?gist=3fddc28e3188fd5f5cded6aa1d037cb0&upload

But when I launch the app from the app launcher, I get:

Execution Interrupted during event processing.
New interpreter error: LOW_MEMORY,MEMORY

Any hints on how I could troubleshoot this?

I have LOW_MEMORY issues when reading my app from storage.
The app is quite big: the code string is 18 156 characters long.

Thank you all.

After I've resolved my DFU issues, I could upload the latest firmware, and it is working as documented.

Geez...
It finally worked. If you don't have your location enabled (on your phone), scan doesn't return anything...

I figured out as it was written in nfConnect tutorial.
I was really worried: I thought I bricked the watch!!

Thanks everyone for your support.

scan does not find it either.

With some screenshots it will help.

Here is the list of bluetooth devices I can reach on my Android phone: DfuTarg is here.

Here the list I have in nfrToolbox, and DfuTarg isnt't available

I've rebooted the phone in case, but it didn't changed.

Sorry if the wording wasn't right.
Using the nRF toolbox, I'm following the procedure, and I'm blocked here:

Tap Select Device and choose the device called DfuTarg

I don't see DfuTarg device in the list my phone displays.

That is why I though I needed to pair.

Hello there, and sorry to re-open the discussion.

When I set my bangle in DFU mode (hold BTN1 + BTN2 until it reboot, and release before the === line is complete), I can see "DfuTarg" bluetooth device in my bluetooth discovery app (laptop and mobile).

When I select it, the watch displays "Connected", and a couple of seconds after, says "Disconnected".
On the laptop/mobile side, the bluetooth app says it could pairing was rejected.

I am pretty annoyed, because my previous attempt to upload a new Firmware failed before completion, and the watch can not boot any more...

Thank you folks.
It must be an outdated firmwire then.
I can not see the current installed version (the boot screen goes to fast :D), and I'll try to apply latest with my Android phone.

I'll let you know.

Hello folks!

First of all, thanks for this amazing job. For a runner like me, having a possibility to make my own sport mode is incredible.

I'm trying to get my heart rate, but Bangle.setHRMPower() function is undefined.
The only example I could find seems to read an analog port instead of listening to the 'HRM' event as the doc says.

I am using the latest firmware (2.04), on the NodeConfEU watch.

Thanks all for the support.
In fact, the firmware was applied, I can tell by the booting screen.

I'm just confused with one thing: on the Android DFU app, I don't have DfuTarg option, but Bangle ECDB instead. Is it ok?

I'm also having some issues, but not exactly the same as Sebi.
My (nodeconf) watch is in DFU started, the exact same screen as you shown, Gordon.

On my Android phone, I select the latest firmware, then my bangle ECDB device, and I hit the upload button (see attached screenshot)

While uploading, nothing happens on the Bangle.
After a while, I'm receiving a notification on the phone (it's too quick to take a screen shot):

Upload failed: GATT ERROR