Implement BLE, using bleak and bless

[WouterJD]

Issue is currently ready for integration in master

By using bless, native python -> BLE libraries are used and the node.js construction can be kept as an alternative.

ToDo

• Testing and verification by others

Advantage

• Easier installation

Risks

• Node.js is a proven solution that works well (after the complex installation); bless and bleak still must be proven to be stable
• The new implementation must/should be tested on windows, raspberry and MacOS

Remark for windows users

• Native bluetooth interfaces cannot be used to advertise BLE (only client devices can be connected)
• When node.js is used, Bluetooth-4 dongles must be used and the driver replaced by Zadig
• When bless is used, Bluetooth-5 dongles must be used and no special drivers are needed

Advertisement
On you smartphone you will see "A smart trainer" and select it...so you might not even notice. The name of the trainer should be "FortiusAnt Trainer".

• When FortiusAnt is running on Windows, the name is the Windows hostname
• When FortiusAnt is running on Raspberry, the name is the bluetooth ADDRESS

No solution yet available from bless; note that such naming differences also exist using -b node.js

---

It's under investigation; see bless and bleak for currently open issues.
bless - the library for a new FortiusAnt BLE interface (BLE server)
bleak - the library to debug FortiusAnt (BLE client, simulating TrainerRoad or Zwift)
kevincar/bless#60 (No access on raspberry pi, bypass available)
kevincar/bless#74 (sample FTMS, under discussion)
kevincar/bless#76 (How to set Generic Access Profile, not blocking, previously issue#75)
not an issue anymore:
hbldh/bleak#688 (timeout issue, wait loop implemented)
hbldh/bleak#775 (Indications not received on Windows 10, no answer, does not occur now)
closed:
kevincar/bless#61
kevincar/bless#63
kevincar/bless#66
kevincar/bless#67 (should be solved)
kevincar/bless#72
kevincar/bless#73; answered
hbldh/bleak#739
hbldh/bleak#782; DeviceName and attributes
hbldh/bleak#772

[WouterJD]

I'm working on this issue for a native python implementation for BLE; to remove the nodejs complexity AND to resolve issues (the structure is too complex).

First version is in https://github.com/WouterJD/FortiusANT/tree/Implement-BLE-using-bless

cleCollector.py is a client to any FitnessMachine (trainer, FortiusAnt) like TrainerRoad and/or Zwift with purpose to show data and make some simple simulations.

Although at 98% I do not get it work and have raised an issue with the bleak developer
hbldh/bleak#772 (comment)

Who can tell me what I'm doing wrong 🙂

[decodeais]

I tried to understand what you are doing. I have not the background and the knowledge to help you. But why can't you you use a ready BLE Analyser like the one from Keuwlsoft. It reads all messages and had the possibility send something? If the question to stupid please erase this text.

[Toetta]

Im happy to try on my installation if you can harvest any info from it... Otherwise I think everything is way beyond my level

[WouterJD]

@decodeais not a stupid remark at all. When I learn an interface, I tend to build something useful. In doing it, I understand the issues, also for other programmers.
So it's not so much BLE-analysis but more end2end interface building. This time for bleak.

[MeanHat]

@WouterJD - I remember you saying that you wanted to simplify the BLE installation - that's great work to remove node.js. I've no expertise in this area but would be very happy to lend a hand with RPi testing (0W, 02W, 3B+) and the manual if you want any assistance. Keith

[WouterJD]

@MeanHat thanks. First client version will be available this week. This client can be used to "see" (and Analyse) what is transmitted by FortiusAnt.

[MeanHat]

Great work @WouterJD !

[Toetta]

👍🏻🥂🤘🏻🔥

[WouterJD]

https://github.com/WouterJD/FortiusANT/blob/Implement-BLE-using-bless/pythoncode/bleClient.py
is now available.
You can use this program to validate the transmission of FortiusAnt (BLE).
The program works on raspberry rpi0W raspbian (10) buster
It does not work well on Windows 10 (indications not received), reason unknown

• Start FortiusAnt -b -s (bluetooth, simulation mode)
• Start bleClient.py (no parameters required)

See also: hbldh/bleak#772 (comment)

[WouterJD]

Good news, prototype for bleServer.py is working and pairing with Trainer Road and Rouvy.
To be cleaned for testable version...
Work for next week :-)

[MeanHat]

Good news, prototype for bleServer.py is working and pairing with Trainer Road and Rouvy. To be cleaned for testable version... Work for next week :-)

That's really great news @WouterJD - great work! That is a gamechanger! I'd be delighted to test with Zwift and Fortius when ready. On standby...

[decodeais]

Good news, prototype for bleServer.py is working and pairing with Trainer Road and Rouvy. To be cleaned for testable version... Work for next week :-)

I would like to test pre alpha state to. It would be interesting to get a chance for bug hunting.

[WouterJD]

It's like a cliffhanger... To be continued in next episode😀

[MeanHat]

Can't wait!

…

On Fri, 25 Feb 2022, 17:13 Wouter Dubbeldam, ***@***.***> wrote: It's like a cliffhanger... To be continued in next episode😀 — Reply to this email directly, view it on GitHub <#366 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ARS665FOXR6YLFPGFC27S7LU462CTANCNFSM5MIAN3TA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[BikeBeppe64]

It's like a cliffhanger... To be continued in next episode😀

All we need is.... your next episode!!

[MeanHat]

Hi @WouterJD - I've been testing today as you requested (using Pi 0 2 W as it is a lot quicker and I can then revert to Pi 0 W). I have done / found the following:

1. There was an error cloning the branch. I increased the swapfile size from 100 (to 1024) to allow me to clone the branch. sudo dphys swapfileoff; sudo nano /etc/dphys-swapfile changing to CONF_SWAPSIZE = 1024; sudo dphys-swapfile setup; sudo dphys-swapfile swapon
2. I modified FortiusAnt.sh to -a -bb -D-1 -s -d CEWIDbu
3. There was an error finding lib_detect_testenv so I installed manually git clone https://github.com/bitranox/lib_detect_testenv.git followed by cd lib_detect_testenv and sudo python setup.py install
4. There was a service_identity error so I installed manually python3 -m pip install service_identity
5. I am now getting an error: "org.freedesktop.DBus.Error.AccessDenied: Connection 1.10 not allowed". I ran sudo usermod -G bluetooth -a pi and this resulted in the same error but Connection 1.49

The log file is attached for info: FortiusAnt.2022-03-04 16-25-06.log

I've had to stop at this stage but happy to pick up again at a future date. Hope this is a small help.

[WouterJD]

Hi guys, I'm still engineering the code. It's not trivial.
As you see in the headers, some questions and support required from bless & bleak :-)

[WouterJD]

5. I am now getting an error: "org.freedesktop.DBus.Error.AccessDenied: Connection 1.10 not allowed". I ran sudo usermod -G bluetooth -a pi and this resulted in the same error but Connection 1.49

See bless/issue/#60

In
/etc/dbus-1/system.d/bluetooth.conf
add

  <policy user="pi">
    <allow own="org.bluez.FortiusAntTrainer"/>
    <allow send_destination="org.bluez.FortiusAntTrainer"/>
  </policy> 

[MeanHat]

Great work @WouterJD! I should have some time for more testing to support you next week and will run the addition you suggest.

[WouterJD]

Current version is 99% final.
There is one open issue; the DeviceName and Appearance cannot be set. Let's hope @kevincar can answer how to do that. It's not blocking for TrainerRoad or Rouvy.

[Toetta]

Fantastic news, great work

[WouterJD]

Fantastic news, great work

Hi Malkus, feel free to test!

[WouterJD]

When FortiusAnt is advertising over BLE, depending on the interface and backends, the CTP (Trainer Road, Rouvy) will see the following devices:

a	Windows	Raspberry
-b using NodeJs	hostname	FortiusANT trainer
-bb using Bless	hostname	BLE-ADDRESS
Useful sources:		check here and here	It did not help me

With my knowledge and what I found, there is not much I can do about it.

There is no practical difference, since your CTP sees one usable device and you select it.
Having "FortiusAnt Trainer" as displayed name would have been more elegant.

For I now, I surrender.

[Toetta]

64bit Bullseye up and running with FortiusANT using bless =)

[WouterJD]

Congrats. Does your app recognize it?

[Toetta]

Yes! Have taken a ride with Zwift and worked well!

Possibly dropping cconnection adter some time. I will have to investigate that further whats going on

[WouterJD]

Cheerio! Let me kno😀

[MeanHat]

Hi @WouterJD - I've been trying bless branch on RPi0 2W. Clean reinstall with Debian Bullseye 32bit.

• I had to change the permissions on some of the new raspberry *.sh files
• I had to install lib_detect_testenv using the install I described above https://github.com/WouterJD/FortiusANT/issues/366#issuecomment-1059339128 and also increase the swapfile size to 1024 to prevent error when using git clone to get the new branch "Implement-BLE-using-bless".
• I used options -a -bb -s -D-1 -d CEWIDbpjmuafA in FortiusAntMain.sh
• FortiusANT ran yet while using bluetooth on CTP=Zwift running on either Win11 or Android devices I could not detect any bluetooth devices
• I have attached the logfile:
  FortiusAnt.2022-03-30 16-24-32.log
• I have no doubt made a mistake somewhere. Do you have any suggestions?

[Toetta]

I had problems getting Zwift to have a stable connection with the 64 bit. The raspberry disappeared from the available devices and got lost signal responce from zwift. Possibly resolved just now by installing the x11-utils package..? Now managed to pedal a couple of km with stable connection :) will continue later

[WouterJD]

Hi @WouterJD - I've been trying bless branch on RPi0 2W. Clean reinstall with Debian Bullseye 32bit.

• I had to change the permissions on some of the new raspberry *.sh files
  Thanks

• I had to install lib_detect_testenv using the install I described above https://github.com/WouterJD/FortiusANT/issues/366#issuecomment-1059339128 and also increase the swapfile size to 1024 to prevent error when using git clone to get the new branch "Implement-BLE-using-bless".
  Is this bullseye specific, since I use the exact same Raspberry installation as before and not required

• I used options -a -bb -s -D-1 -d CEWIDbpjmuafA in FortiusAntMain.sh
  There is no need to modify FortiusAntMain, usually parameters would be set in FortiusAnt.sh, or is that what you mean?

• FortiusANT ran yet while using bluetooth on CTP=Zwift running on either Win11 or Android devices I could not detect any bluetooth devices
  Did you try bleBleak.py to act as a testing client?

• I have attached the logfile:
  FortiusAnt.2022-03-30 16-24-32.log
  I see no special issues; other than that there is not very much of logging by bless

• I have no doubt made a mistake somewhere. Do you have any suggestions?
  Hmmm... not right now

[WouterJD]

I have uploaded some error-handling improvements and access rights settings

[WouterJD]

• I used options -a -bb -s -D-1 -d CEWIDbpjmuafA in FortiusAntMain.sh

To avoid this, you can use -d all or -d log; the difference is that log will not create a json-file.

Note that in the standard python logging system, the messges are hierarchical:

• D(ebug) produces D, I, W, E, C
• I(NFO) produces I, W, E, C
• W(ARNING) produces W, E, C
• E(RROR) produces E, C
• C(RITICAL) produces Critial messages only

-d CEWID is therefore identical to -d D

As opposed to the FortiusAnt flag-approach, where -d DW would produce Debug and Warning but not the others.

[MeanHat]

In answer to your points above #366 (comment):

• OK
• I suspect it might be Bullseye specific
• My mistake - you are correct. I meant FortiusAnt.sh
• Yes, I tried running bleBleak.py but without success
• OK
• OK

Testing again...

[WouterJD]

I tried running bleBleak.py but without success

Please explain, copy output...

[MeanHat]

Here's the bleBLEAK.sh output:
bleBLEAK_output-22-04-14.txt

[MeanHat]

Hi @WouterJD - I'm testing again, this time using Debian Buster rather than Bullseye on an RPi0 2 W and here is an update of the results:

• in 3_InstallBless_Temporarily.sh: (a) line 3 there is a semicolon ";" which causes an error - I replaced with hyphen "-" which has removed error; (b) error reported stating need python >=3.6 and only v2.x available. In line 4 I replaced "pip" with "pip3" which removed the error; (c) Path error so added home/pi/.local/bin to PATH
• ran FortiusANT.sh with -a -bb -s -D-1 but Zwift did not find any device
• ran Test bleBleak.sh but MAC address of Android device running Zwift (Google Pixel 4a 5G) not found

I will do a fresh install of Buster and try Node.js to check "original" setup is still functioning and problem is not related to RPi0 2 W

[MeanHat]

ps - I've been using RPi0 2 W rather than RPi0 as it is a lot faster to set up etc.

[MeanHat]

RPi 0 2 W worked with Node.js (though needs fix in issue #57

Will now test RPi 0 with Bullseye and bless

[WouterJD]

Special.
I've updated 3_InstallBless_Temporarily.sh (to be published)
Will communicated what device I have, I think it's a new type because booting much faster than I was used to.

[MeanHat]

Here's an update on the RPi testing:

OS	Buster 10.12	Buster 10.12	Bullseye 11.3	Bullseye 11.3
BLE Mode	Node.js	bless	Node.js	bless
RPi0	OK (other than disconnect - see issue #57)	NO (see log)	OK (other than disconnect - see issue #57) (see log)	NO (see log)
RPi0 2	OK (other than disconnect - see issue #57)	NO (see log posted previously)	NO	NO (see previous post)

I have focussed on RPi0 as RPi0 2 is fairly different (ARMv7-l vs ARMv6-l, Bluetooth 4.2 vs 4.1 etc). The closest I came was RPi0 running Buster and bless linking to Zwift as the CTP. Zwift recognised FortiusANT but could not achieve a stable connection in simulation mode. Hopefully the log file will help. During the install (running shell scripts 1, 2, 3_InstallBless_Temporarily.sh, 5, 9_GrantAccessToBluetoothForBless.sh and 9_GrantAccessToUSB_withReboot.sh the main errors were (a) in shell 3 I had to change line 4 from python to python3 and (b) there was a prompt to add /home/pi/.local/bin to PATH.

Hope this helps a little. I'm keen to carry on the work to get RPi0 running Buster working with bless. (Next steps could then be to extend the solution further to use Bullseye and then RPi0 2?)

RPi0_bullseye_bless.log
RPi0_bullseye_nodejs.log
RPi0_buster_bless.log

[MeanHat]

Special. I've updated 3_InstallBless_Temporarily.sh (to be published) Will communicated what device I have, I think it's a new type because booting much faster than I was used to.

Thank you - I used this in the testing reported above

[WouterJD]

Implemented and released

[WouterJD]

Here's an update on the RPi testing:

Hi @MeanHat I have released the BLE/bless version today, and I come across this test-work in my Inbox-to-be-handled.
Please share your thoughts.

I'm trying to get clean for the tacx-season :-)

[MeanHat]

Hi @WouterJD - I am sorry for the slow reply. I've been working on some other projects and got distracted! I have not tested bleak & bless since my last trials (I submitted feedback and comments). If you'd like me to test anything again on RPi3, RPi0 or RPi0 2 please let me know.

[WouterJD]

Hi @MeanHat
I did not do a full regression test, so such a test would be very welcome.
If there are any issues, we can quickly resolve.

I wanted to proceed with steering, using the BLE base, without keeping branches open - although technically possible

[MeanHat]

Ok @WouterJD - leave it with me for a few days and I'll do a full test.

[WouterJD]

👍

[WouterJD]

Have been quite busy lately; just stepped in again and did some small stuff.
RPi testing will come later.

Now trying to get TrainerRoad back up again.