HRV data missing in XDATA

[Mario Pabst]

Hello,

I'm new here so please forgive if I'm not following the right procedure or did not search deep enought to find an asnwer to my issue....

I'm using a Garmin HRM Pro strap and an ANT+ USB dongle to read in HR and RR/HRV data into GC.  This works fine with the exception that sometimes the HRV colum containing the RR values is missing completely. 

I was trying to find a correlation when the HRV data is missing, e.g. when swapping to another window app while recording, but can't find any.

Using W10 Laptop and have tried GC 3.5 and 3.6

And sometimes the HRV column is present, but only the last RR value is written.....

I will attache a PDF with my notes taken during the measurements.

Anyone have a clue what causes the issue and what could be done to prevent it?

thanks

mario

   

[Ale Martinez]

There is no additional configuration, but ANT+ messages are received in real time without retransmission, they can be lost due to interference (see https://github.com/GoldenCheetah/GoldenCheetah/wiki/ANT--and-BTLE-Troubleshooting#data-spikes-and-dropped-connections) or missed if the receiver is not ready (may be your hardware is not fast enough for multitasking while training)

[Mario Pabst]

Ok, thanks for the hint.

Re interference, receiver and transmitter have been very close together during my testing, less than half a meter I'd say.

I will close all other apps on my PC and do some more testing to make sure there is enough reccourses left.

B.t.w a friend of mine is having exactly the same issue on his PC, same sensor.  I was really wondering why I could not find this issue being reported already.....

[Mario Pabst]

What I did: Power cycle PC and just started GC.  with no other apps open, I did about 25 records, appr. 10 sec each and of those I got 4 records where the RR file is not written into 
...AppData\Local\GoldenCheetah\<Athlete>\records
On the last 10 records I had the Task Manager open, CPU load averages at 3..5%, RAM is at a constant 30% (11GB free)

To me this doesn't look like a connection or resource issue, unfortunately.

 

I'm working on a little project using GC to provide training guidance during workouts and the RR value is part of it.  If its not recorded, the workout would need to be repeated… not so nice.

 

I wonder if I can put the RR reading provided by the HRM pro directly into the live train view ?  This way one could see right away if it is present/valid when starting the workout.

Have watched some videos, can manipulate the existing train view charts but implementing RR seems more tricky, right?

[Ale Martinez]

El martes, 25 de enero de 2022 a la(s) 10:49:11 UTC-3, gmap...@gmail.com escribió:

What I did: Power cycle PC and just started GC.  with no other apps open, I did about 25 records, appr. 10 sec each and of those I got 4 records where the RR file is not written into 
...AppData\Local\GoldenCheetah\<Athlete>\records
On the last 10 records I had the Task Manager open, CPU load averages at 3..5%, RAM is at a constant 30% (11GB free)

To me this doesn't look like a connection or resource issue, unfortunately.

Processing of RR messages is very simple: when the first message arrives the records/*.rr file is created and appended in buffered mode for each subsequent one, caching is flushed and file closed when the training session stops: https://github.com/GoldenCheetah/GoldenCheetah/blob/0449988fb1f71b690b75ed4cf7e03993f92a9427/src/Train/TrainSidebar.cpp#L3112

So, if the file is not even present, no RR message was received for that session.

Have you tried recording using other app on the same hardware or even using a Garmin or similar watch to discard a faulty strap? 

I'm working on a little project using GC to provide training guidance during workouts and the RR value is part of it.  If its not recorded, the workout would need to be repeated… not so nice.

 

I wonder if I can put the RR reading provided by the HRM pro directly into the live train view ?  This way one could see right away if it is present/valid when starting the workout.

Have watched some videos, can manipulate the existing train view charts but implementing RR seems more tricky, right?

RR data is not currently available for realtime charts or dials. 

[Mario Pabst]

Thanks Ale for your help.

I have used the Heart Monitor+HRV App from f3b on a Garmin Venu 2s and it was working without any issues, never missed any data.

I don't have the watch anymore.  With the connect app from Garmin on my iPhone I'm not sure how to test the rr recording.

And nor I'd know how to use another app on my PC.  Any idea?

Also, my colleage is using the same strap on another W10 PC and seeing exactly the same issues....

[Ale Martinez]

El martes, 25 de enero de 2022 a la(s) 18:47:48 UTC-3, gmap...@gmail.com escribió:

Thanks Ale for your help.

I have used the Heart Monitor+HRV App from f3b on a Garmin Venu 2s and it was working without any issues, never missed any data.

I don't have the watch anymore.  With the connect app from Garmin on my iPhone I'm not sure how to test the rr recording.

And nor I'd know how to use another app on my PC.  Any idea?

Also, my colleage is using the same strap on another W10 PC and seeing exactly the same issues....

The USB ANT+ sticks are also Garmin?

There are known issues with others: https://github.com/GoldenCheetah/GoldenCheetah/wiki/ANT--and-BTLE-Troubleshooting#cycplusanself-ant-stick

[Mario Pabst]

Yes, Garmin ANT+ sticks.

I will order a new stick and another strap, HRM-Run most likely, just to exclude those from the list.

[roxystar]

First of all: GoldenCheetah is a unique SW which i do like very much and use for several years now. Thank you for the passion and dedication for such a long time to keep this moving.

I can share the same observation on the HRV recording during trainer rides.

Running GoldenCheetah on Lubuntu 21.04

Ant+ stick: Garmin Dynastream

Chestbelts tested: Polar H10, BerryKing, Garmin

all show the same above described behaviour

my workarounds currently are

(1) run fatmaxxer app on Lenovo Duet tablet in parallen with Polar H10 and do post ride activity combination (sync against heartrate readings of GoldenCheetah), advantage: i can see real time DFA alpha 1 values on the tablet

(2) run HRV+ app on old iPhone in parallel with BerryKing and do post ride activity combination

In the beginning i did several seconds of train mode in order to see if HRV are recorded - however i gave up because i was not able to find out what systemic influence allowed for recording or not. I tested with a 10 min warm up ride before (to get a measure for current fatigue based on DFA alpha1) and than the real ride. Sometimes both rides do have HRV, sometimes only the first and sometimes only the second.

Experiments: Switch of all WLAN, switch off BLE devices in the room, get ANT+ stick close to sender by USB extension cable

Again thank you for all the great work!

Robert

[Mario Pabst]

Hi Robert,

thanks for supporting me on this  ;-)  Thought I'm doing smth wrong here.

I'll get my new strap and new ANT+ dongle tomorrow but after reading your post I have little hope that it'll be the solution.

Looks like this function is not used widely here, unfortunately

-mario

[Mark Liversedge]

Hi

ANT+ HRV Data is recorded during the workout if the cheststrap supports ANT+ recording of HRV data.

If you use a cheststrap that does not send RR data over ANT+ then it will not be recorded, because it was never sent.

You can easily validate if HRV data was recorded/sent by looking for the "rr" file as described by Ale higher up in this thread.

The RR data is appended as XDATA when you import the recording (popup when the workout ends) and will be shown in an "HRV" tab on the ride data editor.

Mark

[Mario Pabst]

Hi Mark,

thanks for for your comments and much appreciated your guys time and devotion to this fantastic piece of software and to OUR issues!

Yes, all understood. 

The strap (HRM Pro) supports ANT+ recording of HRV.  In about 80..85% of the recordings HRV is recorded together with HR in GC and generated files into the record folder.

The question is: What is happeneing on the 20% where it is not recording HRV but only HR.

When I used the very same strap together with a Venu 2s and some HRV apps it was always providing HRV numbers.

So I have little doubt that the issue is hardware related.  a) as a friend of mine is seeing the same issue with another HRM Pro on another PC and b) as Robert is reporting the same issue with several straps ...

Atatched a screenshot of the records folder where you can see the missing .rr files in comparison with all existing .csv files, which get written consistently with each record.

Side issue:
Besides the fact that sometimes (15..20%) the entire .rr file is not written, in about the same amount of cases the .rr file is written, but GC just displays the last value of the data series. Initially the rr file does contain only the last RR value. BUT, after taking another recording, and the next csv and rr files are written, the initial rr file then will be completed with all rr data.

However, GC will keep displaying only the last rr value, even though the rr file now contains all rr numbers.  See 2nd screenshot.
ANd as I write this, I see the last value beeing the first in the rr file, and then the other values with time stamps starting from zero... an indication of the writing order  

thanks so much,

mario

[Mario Pabst]

...received another Garmin strap (HRM dual) and a new Garmin ANT+ USB stick: Same behavior... some recordings do not write the rr file.

[Mark Liversedge]

On Thursday, 27 January 2022 at 16:09:05 UTC gmap...@gmail.com wrote:

Side issue:
Besides the fact that sometimes (15..20%) the entire .rr file is not written, in about the same amount of cases the .rr file is written, but GC just displays the last value of the data series. Initially the rr file does contain only the last RR value. BUT, after taking another recording, and the next csv and rr files are written, the initial rr file then will be completed with all rr data.

However, GC will keep displaying only the last rr value, even though the rr file now contains all rr numbers.  See 2nd screenshot.

This looks like a recording bug

 

ANd as I write this, I see the last value beeing the first in the rr file, and then the other values with time stamps starting from zero... an indication of the writing order 

This is a bug too, just because time goes backwards, we should not be discarding data.

Mark

[Jozef]

Hello and sorry for the intrusion in the discussion for an off topic request. Currently for the measurement of HRV I use the Italian software TRAIN ME COACH by Hosand with now obsolete minicardium. The software stores data from HRV bands but before buying one I wanted confirmation that the HRV trace is actually being read. For this I ask your willingness to provide me with a file containing HRV data to allow me to verify that the software actually reads it? From now on, I thank those who will be available to help me.

[Ale Martinez]

I am wondering about a concurrency problem since the rrFile object is created and appended by ANT+ events, but closed by a GUI event, should we guard it using a QMutex?

BTW, if this is the case VO2max recording should have similar problems, but not standard recording since the timer is stopped before closing the file. 

[Mark Liversedge]

On Friday, 28 January 2022 at 13:21:06 UTC Ale Martinez wrote:

I am wondering about a concurrency problem since the rrFile object is created and appended by ANT+ events, but closed by a GUI event, should we guard it using a QMutex?

My feeling is the use of signals is the problem as they span threads and get queued, but not really sure ?

Mark

[Ale Martinez]

That's a better explanation for the delayed writes, but harder to fix?

BTW, creating and destroying a QTextStream for each RR can be expensive due to the flush to disk at destroy time, we could keep it alive to benefit from the buffering, and destroy it only before to close the file.

[Mark Liversedge]

I went to try and work on this but my Garmin HRM is dead and as usual the little screws to replace the battery are screwed tight but have nice soft steel to sheer unless you have the right micro screwdriver to remove. I know it's off-topic: but please-- avoid Garmin HRM they are a scam (cue someone saying its to make them hermetically sealed whilst ignoring the fact the screws are deliberately made of soft steel).

Anyway, back on topic, I think we should just add the R-R data to the realtime data class and write to the ".rr" file-- and consider moving to the CSV file in the future, along with adding the R-R data series to ridefilepoint too.

Mark

Mark

Mark

[Mario Pabst]

Anyway, back on topic, I think we should just add the R-R data to the realtime data class and write to the ".rr" file-- and consider moving to the CSV file in the future, along with adding the R-R data series to ridefilepoint too.

This sounds like a bigger thing and requiring  the cover of a release?

[Ale Martinez]

El lunes, 31 de enero de 2022 a la(s) 08:24:10 UTC-3, gmap...@gmail.com escribió:

Anyway, back on topic, I think we should just add the R-R data to the realtime data class and write to the ".rr" file-- and consider moving to the CSV file in the future, along with adding the R-R data series to ridefilepoint too.

This sounds like a bigger thing and requiring  the cover of a release?

It is a non trivial change, AFAIK.

The realtime data class only holds the last sample for each series, and for R-R data that is not enough, since there can be multiple R-R messages for the same 1 sec sampling period. 

[Mario Pabst]

Thants what I was afraid off.

It would help me a lot if the writing/not writing would be somewhat predictable. Or the other way:  what do I need to do/not do in order to be sure the rr file is written and complete, at this state of the build. Any idea?  

[Ale Martinez]

El lunes, 31 de enero de 2022 a la(s) 12:21:46 UTC-3, gmap...@gmail.com escribió:

Thants what I was afraid off.

It would help me a lot if the writing/not writing would be somewhat predictable.

I think there can be 2 independent problems: one is ensuring the data reaches GC, for that you likely need to use an USB extender cable to put the ANT+ receptor close enough to the HR band, similar to when you used a watch or cell phone to do the recording.

 

Or the other way:  what do I need to do/not do in order to be sure the rr file is written and complete, at this state of the build. Any idea?  

 The other is delayed messages, this likely can be avoided waiting a while before to import the recorded session.

[Mario Pabst]

Ok, thanks Ale.

For all my testing I was wearing the strap on my chest, sitting directly in front of the laptop, so less than 60.70 cm between strap and ANT+ stick.  

I will get an extender cable and bring the ant+ stick within 10 cm to the strap.

 The other is delayed messages, this likely can be avoided waiting a while before to import the recorded session.

Not sure I understand:  I'm in the train dashboard on my Laptop, have reasonable HR readings and start the recording by hitting PLAY.  10 sec later I hit STOP and the "Importing Files" window pops up.  And before hitting "Save" I should wait a few sec ?  

[Ale Martinez]

El lunes, 31 de enero de 2022 a la(s) 16:48:46 UTC-3, gmap...@gmail.com escribió:

Ok, thanks Ale.

For all my testing I was wearing the strap on my chest, sitting directly in front of the laptop, so less than 60.70 cm between strap and ANT+ stick.  

I will get an extender cable and bring the ant+ stick within 10 cm to the strap.

 The other is delayed messages, this likely can be avoided waiting a while before to import the recorded session.

Not sure I understand:  I'm in the train dashboard on my Laptop, have reasonable HR readings and start the recording by hitting PLAY.  10 sec later I hit STOP and the "Importing Files" window pops up.  And before hitting "Save" I should wait a few sec ?  

No, what I mean is: once your training has finished don't stop it immediately, so you let time for the pending signals to drain and be recorded to the .rr file, after a while stop the workout and import it. You can easily delete the unwanted samples at the end using Editor afterwards.  

[Mario Pabst]

ok, understand what you mean.

However, the point is the rr file is either written for a given recording, or not. And if not, it does not help starting the recording earlier before the training and let it run longer after the training. If it is not there, its not there.

 

Just for clarification: I'm trying to monitor HR during an exercise and like to record HR and HRV for post exercise evaluation.

I’m wearing the strap and have the laptop with ANT+ stick close by. No other hardware.

I open GC, go to my TRAIN dashboard, start the workout by pressing the red button, which turns green.  HR numbers from the strap look reasonable on the dashborad.  So I start the recording by pressing the PLAY button.  Then doing my exercise.  I double check HR which is fluctuation, and not stuck….
I stop the exercise by hitting the stop button on the dashboard, then hit save and finish on the two popup windows.
Then go to Activities and check if HR and HRV are present for the just recorded data set. And sometimes the HRV is just missing or just the last value has been written, as described before.

 And I did some more testing with the ANT+ stick really close to the Garmin HRM Pro (less than 10 cm) and let some time pass after stopping the recording, before hitting SAVE on the popup. No difference, unfortunately.

[Ale Martinez]

El martes, 1 de febrero de 2022 a la(s) 07:25:40 UTC-3, gmap...@gmail.com escribió:

ok, understand what you mean.

However, the point is the rr file is either written for a given recording, or not. And if not, it does not help starting the recording earlier before the training and let it run longer after the training. If it is not there, its not there.

my hypotheses was these cases were due to transmission problems 

Just for clarification: I'm trying to monitor HR during an exercise and like to record HR and HRV for post exercise evaluation.

I’m wearing the strap and have the laptop with ANT+ stick close by. No other hardware.

I open GC, go to my TRAIN dashboard, start the workout by pressing the red button, which turns green.  HR numbers from the strap look reasonable on the dashborad.  So I start the recording by pressing the PLAY button.  Then doing my exercise.  I double check HR which is fluctuation, and not stuck….
I stop the exercise by hitting the stop button on the dashboard, then hit save and finish on the two popup windows.
Then go to Activities and check if HR and HRV are present for the just recorded data set. And sometimes the HRV is just missing or just the last value has been written, as described before.

 And I did some more testing with the ANT+ stick really close to the Garmin HRM Pro (less than 10 cm) and let some time pass after stopping the recording, before hitting SAVE on the popup. No difference, unfortunately.

i don’t have a band for testing and run out of ideas, so I am finished with this.

May be you can use the workaround Peter describes in a previous post https://groups.google.com/g/golden-cheetah-users/c/wBYltBb5zqo/m/4msAjClWAgAJ

[Ale Martinez]

El martes, 1 de febrero de 2022 a la(s) 07:59:01 UTC-3, Ale Martinez escribió:

May be you can use the workaround Peter describes in a previous post https://groups.google.com/g/golden-cheetah-users/c/wBYltBb5zqo/m/4msAjClWAgAJ

Robert, not Peter, sorry. 

[Mario Pabst]

Yes...thanks Ale, much appreciated.  I'll look for alternatives using other hardware/brands.

[erik...@gmail.com]

I've looked into this a bit by creating a bunch of small recordings trying to reproduce the issue with some extra debug info added, and I've found two things.

1. There is indeed an issue where the TrainSidebar:Stop() close the .rr file, but directly afterwards TrainSidebar::rrData() is called which will then open the file again and truncate it so all previous data is lost.

2. When issue 1 has happened, this also means that the rrFile is already open when the next training session is started which means that the new rrData will be logged into the wrong .rr file and thus not show up in the recording at all.

I suspect that it could be enough to just make sure the RT_RECORDING bit is cleared before closing rrFile, then if any extra calls to rrData() is done they won't be a problem since it won't do anything if RT_RECORDING is not set.

Unfortunately I have to test many times to trigger this bug, so it's hard to tell if it's fixed or just not triggered.

Anyway, I'll probably get some time during the evening or at least weekend to test more and produce a pull request for a suggested fix.

Cheers,

Erik

 

[Ale Martinez]

That’s interesting, Erik, for that to happen rrdata signal needs to be executed under a different thread from the (GUI) Stop event, and in this case clearing status before closing rrfile should help but it may be not enough, since critical race conditions could still arise, we would likely need to guard status with a qMutex to avoid it completely.

[Ale Martinez]

I think to add QMutexLocker to Stop and rrData functions with a member QMutex, would be a simple and clean way to fix this with minimum changes

[erik...@gmail.com]

That sounds like a good plan.

The rrData signal is emitted from the ANT thread, which is indeed it's own thread. I believe Qt should default to using a Queued Connection if the signal is emitted in another thread than the receiver exist in, and then I don't really understand why this still happens. It only makes sense if Qt uses the Direct Connection type. But I still think that using a QMutexLocker/QMutex combo in Stop() and rrData() is preferable to enforcing a Queued Connection, since it protects against the problem rather than removing a case which triggers the problem.

[Ale Martinez]

I pushed a fix for testing in https://github.com/GoldenCheetah/GoldenCheetah/commit/3db1839471bbe4bbf50267c951d0c7375c177a21, once the build is finished a v3.6 installer for Windows can be downloaded from https://ci.appveyor.com/project/Joern-R/goldencheetah-knhd8/builds/42456245/artifacts

[Ilja Booij]

The issue with the ANT Thread not using a QueuedConnection is probably caused by the fact that ANTlocalController::myANTlocal is a QThread, which has the ANTlocalController as it's parent. So myANTLocal lives on the same thread as the ANTlocalController, which is, if I'm not mistaken, the main thread. Because of this, signal-slot connections will be DirectConnections, not QueuedConnections.

Op vrijdag 4 februari 2022 om 13:11:25 UTC+1 schreef erik...@gmail.com:

[Ale Martinez]

El viernes, 4 de febrero de 2022 a la(s) 18:26:54 UTC-3, Ilja Booij escribió:

The issue with the ANT Thread not using a QueuedConnection is probably caused by the fact that ANTlocalController::myANTlocal is a QThread, which has the ANTlocalController as it's parent. So myANTLocal lives on the same thread as the ANTlocalController, which is, if I'm not mistaken, the main thread. Because of this, signal-slot connections will be DirectConnections, not QueuedConnections.

I think this is good, provided we avoid interferences, I made a small change to have separate locks for rrdata (ANT+) and vo2 (BTLE) so they can write their files concurrently as before, we need coordination with the Stop event, but not between them:  https://github.com/GoldenCheetah/GoldenCheetah/commit/a13da9e91751566fec544673b4b2509b12821a3b

[Ale Martinez]

The changes discussed in this thread are included in last release: https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-RC1, it would be interesting to know if they work for the people who reported problems here.

[roxystar]

Ale,

today i tested the new RC in regard to the HRV recordings. I was able to do 4 different mrc rides and all had complete and correct HRV data (compared against fatmaxxer csv files). In the past i had not achieved 4 correct recordings in a row. I will continue testing the next days to get better confidence level.

Thanks a lot for all the effort, i was really happy when i looked at the files and all data were included. This will allow me to continue my experiments around HRV and will make my workflow much more straight forward.

Robert

[Ale Martinez]

Great, thank you for the feedback.

[Mark Liversedge]

On Wednesday, 9 February 2022 at 21:23:56 UTC roxystar wrote:

This will allow me to continue my experiments around HRV and will make my workflow much more straight forward.

Robert

Hi Robert,

Feel free to share your experiments here, I am sure a lot of folks will be interested to hear.

I am particularly interested in adding richer HRV data to support planning (alongside RPE, ROF, Power, HR et al)

Mark

 

[Mario Pabst]

Hi Ale,

thanks a lot for the quick changes you applied. Fantastic!

A qick testing this morning looks quite positive:  No single recording did miss to write the rr file, so HRV data is available for all records!!!  I did about 25 readings from 10 to 120 sec with no single faliure.

However, I did have 3 chrashes of GC during the testing, each time at the import of data which is automatically executed after the stop of the recording.  Had to close and restart GC every time.  The good thing though is: the rr and the csv file do get written on these recordings, just the automated import seems to cause some issues. After the restart the data is available in activities/details/raw data!  So all no recording would need to be repeated.

I have to say that I never had such crashes on V3.5 and V3.6 before, but can't say if it is really related to this patch.

Many thanks,

mario

[Mario Pabst]

forgot the screenshots

[Ale Martinez]

This change has no bearing with import, which has not changed.

Look at the troubleshooting section in FAQs, remove user metrics and create a new athlete to rule out custom chart issues.

[Mario Pabst]

I created a new athlete and didn't change anything from a user perspective other than adding the HRM Pro sensor after the install of this latest version.

But it is still crashing.

Didn't find anything in the FAQ explaining how to make sure that one is using a fresh install and nothing customized. 

Is there a way to reset GC to like a fresh install? 

[Steve Edmonds]

Can I please confirm HRV data is only available through ANT+ and not through a BLE connection to a heart rate monitor.

Recording in the .rr file is working well now with my Garmin strap (ANT+) but not with my Polar strap (BLE). The Polar strap works ok with a HRV app on my phone.

Re. the Check file permissions message. I had that message first train session after upgrade to RC1 (Mac). The files were logged but not imported to activities. Subsequent train sessions were saved and imported without any issues.

[Ale Martinez]

El jueves, 10 de febrero de 2022 a la(s) 17:53:03 UTC-3, gmap...@gmail.com escribió:

I created a new athlete and didn't change anything from a user perspective other than adding the HRM Pro sensor after the install of this latest version.

But it is still crashing.

See the troubleshooting section about crashes when saving, etc.

 

Didn't find anything in the FAQ explaining how to make sure that one is using a fresh install and nothing customized. 

Is there a way to reset GC to like a fresh install? 

See the FAQs on data and look for where data and settings are located 

[Ale Martinez]

El jueves, 10 de febrero de 2022 a la(s) 18:39:25 UTC-3, Steve Edmonds escribió:

Can I please confirm HRV data is only available through ANT+ and not through a BLE connection to a heart rate monitor.

Right.

 

Recording in the .rr file is working well now with my Garmin strap (ANT+) but not with my Polar strap (BLE). The Polar strap works ok with a HRV app on my phone.

Re. the Check file permissions message. I had that message first train session after upgrade to RC1 (Mac). The files were logged but not imported to activities. Subsequent train sessions were saved and imported without any issues.

This is macOS not GC 

[Mario Pabst]

     I created a new athlete and didn't change anything from a user perspective other than adding the HRM Pro sensor after the install of this latest version.

     But it is still crashing.

See the troubleshooting section about crashes when saving, etc.

GoldenCheetah crashes on exit or when saving files · GoldenCheetah/GoldenCheetah Wiki · GitHub

I checked all drivers on my PC related to Display and Graphics, all the latest it could find.

See the FAQs on data and look for where data and settings are located 

Deinstalled this latest GC Dev release

 https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-RC1

renamed the GC folder in C:\Users\<username>\AppData\Local\ to not use old/existing configs

Did a fresh install of the latest GC release. Connected the HRM Pro and did the recordings. Same thing, especially on longer recordings of 1min or more it writes the .json in /activities and the csv and rr into /records but crashes immediately after recording stops and data is attempted to be imported.

Then I tested without the HRM Pro connected. Just starting up GC, start the workout. Then do the recordings. The recordings do not write the rr file in this case.  10 recordings, 10 imports without a single crash.

Then connecting the HRM Pro and do the same recordings.... this time the rr files do get written as well.

1st crash at the 6th recording.

I can live with that, much appreciate fixing the 'intermittant not recodring of HRV data' Ale !!!

Just thought it might be helpful to know that it happens as described.

thanks, 

mario

[Ale Martinez]

After a crash, a crashXXX.log file is generated parallel to athlete’s folders, when GC is started from the file manager instead from a shortcut, the log contains source file names and line numbers which may help to understand where the crash is happening.

[Mario Pabst]

mhhhhh... started GC.exe in its residing folder C:\Golden Cheetah

crash right at the end of the first recording.  But the only  *.log I can find in   C:\Users\<username>\AppData\Local\GoldenCheetah is

goldencheetah.log

and this is still empty. 

[Joachim Kohlhammer]

Hi Ale,

I experienced the same freeze (GC still showing, UI not reacting to any event, no CPU utilization, process was killed manually by me after waiting for some minutes) when saving my training data after a workout. This happened on all three (longer, i.e. > 1h) workouts after the two commits (3db1839,  a13da9e) were applied. For testing purposes I also did two 1 minute trainings, these were saved without a problem by this version.

Today I reverted these commits and saving worked again for a long training.

I did not change any configuration in GC.

Regards

 Joachim

[Ale Martinez]

This freezing happens when you Stop the workout, or when you import the activity afterwards?

Are the corresponding files generated in records folder, and can they be imported after restart?

[Ale Martinez]

El viernes, 11 de febrero de 2022 a la(s) 15:54:01 UTC-3, tiefgara...@gmail.com escribió:

Hi Ale,

I experienced the same freeze (GC still showing, UI not reacting to any event, no CPU utilization, process was killed manually by me after waiting for some minutes) when saving my training data after a workout. This happened on all three (longer, i.e. > 1h) workouts after the two commits (3db1839,  a13da9e) were applied.

Due to the use of QMutexLocker the locks are held while the import dialog is open, this is unnecessary and perhaps it may be the cause of this behavior, I will push a change to hold the locks only when it is strictly necessary. 

[Mario Pabst]

This freezing happens when you Stop the workout, or when you import the activity afterwards?

For me I start the record in Train view,  and then stop it when done.  The import window pops up automatically and then freezes.  The percentage of freezes vs. o correct import is about the same as when before (with previous version) the rr file would not have been written. But that is a total guess.

Are the corresponding files generated in records folder, and can they be imported after restart?

Yes.

[Joachim Kohlhammer]

My workflow is the same as Marios, also the generated files are available in the records folder and can be imported.

I will test your update as soon as it is available.

[Ale Martinez]

I pushed a candidate fix: https://github.com/GoldenCheetah/GoldenCheetah/commit/7417e9cf1f0a129e870e065ee39e395c562c899a, please test if you can build by yourself, or downloading the Windows installer from https://ci.appveyor.com/project/Joern-R/goldencheetah-knhd8/builds/42550526/artifacts once the build finishes.

[roxystar]

Hi Mark,

yes i will do so. It will take me a bit more time as i am just recovering from influenca. I did not enjoy the illness at all however to some degree i like this unplanned "test case". I am very curious what the impact on the different values will be (need to do some data consolidation first).

Robert

[Joachim Kohlhammer]

My first test today was successful, no more freeze! If nothing unexpected happens I should be able to perform two more tests until Wednesday and keep you updated.

[Steve Edmonds]

I ran a 60 minute test yesterday on my Macbook with the unmodified RC1 and although it froze there was only 1 second difference between the .csv and .rr files. There was no issue importing the training session with the .rr data after restarting GC.
Joachim, I would be interested to know what you are doing with your R-R values, are you using a chart to display your analysis.
steve

--
_______________________________________________
Golden-Cheetah-Users mailing list
golden-che...@googlegroups.com
http://groups.google.com/group/golden-cheetah-users?hl=en
---
You received this message because you are subscribed to the Google Groups "golden-cheetah-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golden-cheetah-u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/golden-cheetah-users/e552e028-e598-467c-bcc6-8d852d0bd0b9n%40googlegroups.com.

[Mario Pabst]

Morning

I just installed the latest version (Build date: Feb 12 2022 15:19:53  Build id: 4010) but in TRAIN view, I cannot start the workout by mouse clicking the on/off button at the bottom (red).
Clicking on it does not do anything.

I have to say that with the previous version (from Feb 10th I think) I had an issue with it already, when clicking it mostly did a double click, so ON and right OFF again, so I had to click a couple times really fast to get it to ON, eventually.

 
Now with the latest I cannot switch it ON.

Is there a key combination to get the workout starting?

thanks

mario

[roxystar]

Maybe the root cause is: There is no entry in Equipment/Geräte which is connected.

Robert

[Mario Pabst]

Yes, of course. Rookie mistake. Thanks

[Mario Pabst]

Looks really good so far, did about 25 recordings without a freeze!  Thanks so much Ale!!!

[Joachim Kohlhammer]

My additional 2 trainings (>1h) worked without a freeze. Ale, thank you very much for this fix (and of course all the rest!)

[Mario Pabst]

I pushed a candidate fix: https://github.com/GoldenCheetah/GoldenCheetah/commit/7417e9cf1f0a129e870e065ee39e395c562c899a, please test if you can build by yourself, or downloading the Windows installer from https://ci.appveyor.com/project/Joern-R/goldencheetah-knhd8/builds/42550526/artifacts once the build finishes.

May I ask if there is already a MAC-Version available as well?

[Ale Martinez]

El lunes, 28 de febrero de 2022 a la(s) 09:33:06 UTC-3, gmap...@gmail.com escribió:

I pushed a candidate fix: https://github.com/GoldenCheetah/GoldenCheetah/commit/7417e9cf1f0a129e870e065ee39e395c562c899a, please test if you can build by yourself, or downloading the Windows installer from https://ci.appveyor.com/project/Joern-R/goldencheetah-knhd8/builds/42550526/artifacts once the build finishes.

May I ask if there is already a MAC-Version available as well?

There are new snapshot build including the fix to this issue available for download from http://goldencheetah-binaries.s3-website-us-east-1.amazonaws.com/ 

[Mario Pabst]

Thank you Ale !!!