VERSION 3.5 DEVELOPMENT BUILD 1802

[Mark Liversedge]

There is a new development build available, which includes Xert integration, Python Charts and more HRV and Measures support.

Its available here: https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.5-DEV1802

For more info on Python see: https://github.com/GoldenCheetah/GoldenCheetah/wiki/UG_Special-Topics_Working-with-Python

As ever, please report build issues here.

Mark

[David DePiano]

Please forgive the cross post, but I've deleted everything out of athlete-private.ini. I add a Strava account, click authorize, gain the Oauth token, no problem as my ride downloads. I attempt to add the Withings account in the same fashion, but clicking Authorize results in no box opening to grant access. I have gone through the Withings API documentation and everything appears to be working fine on their end generating Oauth tokens to access my weight data. I'd like to help troubleshoot, but unsure how to help. I followed the steps here https://developer.health.nokia.com/api#step1 and at the end was able to get a successful API response with my weight data. I tried putting my own withings_token and withings_secret in the athlete-private.ini file, but I'm assuming since I don't have access to the Golden Cheetah consumer token and secret that's why I am getting an OAuth invalid message.

What can I provide to help?

[Will]

Feature list looks pretty sweet!

Thanks,
Will

[TommyRaton]

1st attempt to build GoldenCheetah (actually 1st attempt to build anything!)

LinuxMint 18.3 (Ubuntu16.04)

QMake version 2.01a

Using Qt version 4.8.7 in /usr/lib/x86_64-linux-gnu

So what have I forgotten to install or configure as the make fails thusly

-I../qtsolutions/qwtcurve -I../kqoauth -I. -o RideMapWindow.o Charts/RideMapWindow.cpp

Charts/RideMapWindow.cpp: In member function ‘void RideMapWindow::osmCustomTSURLEditingFinished()’:

Charts/RideMapWindow.cpp:266:24: error: ‘class QComboBox’ has no member named ‘currentData’

     switch (tileCombo->currentData().toInt())

                        ^

Makefile:2737: recipe for target 'RideMapWindow.o' failed

make: *** [RideMapWindow.o] Error 1

thanks

[Mark Liversedge]

This is the wrong place to post, but use QT 5.9.x

[Mark Liversedge]

There is a regression, Joern is looking into it.

[TommyRaton]

Where/which is the correct place to post?

[Mark Liversedge]

https://groups.google.com/forum/#!forum/golden-cheetah-developers

[Mark Smith]

I tried the example cadence and power plot from the Wiki and get the following error message in the Python console:

File "<string>", line 9, in <module>
NameError: name 'GC' is not defined

Any ideas why this isn't working. I am on Win 64 Python 3.6.4.

This is the Python script:

import numpy as np
import plotly
from plotly.graph_objs import Scatter, Layout

## Retrieve power and cadence
xx = np.asarray(GC.series(GC.SERIES_WATTS))
yy = np.asarray(GC.series(GC.SERIES_CAD))

## Prepare Plot
f =plotly.offline.plot({
    "data": [Scatter(x=xx, y=yy, mode = 'markers')],
    "layout": Layout(title="Power / Cadence")
}, auto_open = False)

## Load Plot
GC.webpage(f)

[Erik Botö]

On Sunday, February 18, 2018 at 6:17:50 AM UTC+1, Mark Smith wrote:

I tried the example cadence and power plot from the Wiki and get the following error message in the Python console:

File "<string>", line 9, in <module>
NameError: name 'GC' is not defined

Any ideas why this isn't working. I am on Win 64 Python 3.6.4.

I get the same error whenever I try to use GC in python on Mac too. High Sierra + python 3.6.4.

Cheers,

Erik

[Mick Drake]

Hi Mark
I have an issue with synchronising my dropbox.. GC states that the files are already there and no files are listed in the Synchronise screen.  I can go to upload / download and manually select the files and tick over write and then they are downloaded / uploaded.  Noticed because I have a laptop downstairs to record all my rowing stuff and one upstairs connected to my TACX turbo. GC previous build is on PC upstairs and no issues synchronising that one.
Thanks
Mick

[Stefan Rothbauer]

Hi Mark,

on my windows build, after installing v3.6-DEV1802 all R Charts do not work. 

It does not work to add a new R Chart either.

R has been installed with GoldenCheetah and the path in the options is set correctly.

Any ideas on what I might miss?

Any way to recover the R-Code from the charts?

Thanks,

Stefan

[Stefan Rothbauer]

Problem solved. Somehow GC disabled R in the options. Enabling R via the checkbox helps :-)

[Stefan Rothbauer]

Hi,

I am really looking forward to the python charts.

If there will be some more documentation or a video-tutorial on getting one started with the python charts it would be great if it covered also the installation of the dependencies (python modules) or at least a pointer on what one has to do.

I am using python under windows and numpy, pandas etc. are all installed and working, but in GC the python console will not import numpy :-(

Thanks,

Stefan

Am Freitag, 16. Februar 2018 09:44:04 UTC+1 schrieb Mark Liversedge:

[Stefan Rothbauer]

Hi, 

solved the module import problem (different versions of python in windows).

However now in the Python Console I get the Error Message shon in the attachement.

I used the example from the UG. What do I need to do so that the GC functions become available?

Thanks,

Stefan

##

## Python program will run on selection.

##

import numpy as np

import plotly

from plotly.graph_objs import Scatter, Layout

## Retrieve power and cadence

xx = np.asarray(GC.series(GC.SERIES_WATTS))

yy = np.asarray(GC.series(GC.SERIES_CAD))

## Prepare Plot

f =plotly.offline.plot({

"data": [Scatter(x=xx, y=yy, mode = 'markers')],

"layout": Layout(title="Power / Cadence")

}, auto_open = False)

## Load Plot

GC.webpage(f)

[Mark Liversedge]

So make sure you have:
Python3
numpy
plotly

Python3 in path or PYTHONHOME set in environment or preferences.

It works for me.

Mark

[Mark Liversedge]

Is your dropbox folder configured correctly?

[Ale Martinez]

A quick note on Body and Hrv Measures: this version allows to plot daily measures in Metrics Trends Charts independent of the presence of activities for the day, this is useful when Body or Hrv Measures are downloaded or imported from external sources and/or maintained on Athlete's preferences.

The feature is available as a new option in curve settings:

The Hrv Measures chart on the cloud was recently updated to use it.

[Mick Drake]

Working today.. who know what the issue was .. I did restart yesterday but had the same error

[Mick Drake]

Error today with new build and my withings account

{"status":342,"error":"The signature (using Oauth) is invalid"}

On Friday, 16 February 2018 08:44:04 UTC, Mark Liversedge wrote:

[Stefan Rothbauer]

One step further.

I was missing the SIP module. This could be listed in the dependencies section of UG_Special Topics_Working with Python.

PYTHONHOME refers to the field in Options/General (there it is spelled Python Home)? -> that is set.

PythonPath variable is set in windows (Win7).

Still I cannot produce the sample plot.

The error message now indicates it cannot open the temp-plot.html file.

Where would plotly integrated into GC put the generated html files? Just to check if these are produced.

plotly is found and imported.

Any other comments on what to check?

Best, 

Stefan

[Stefan Rothbauer]

Again one step further. plotly seems to have a problem on my system to produce the html files of the plots in whatever is the default location it assumes in its GC integration.

Specifying an output file (filename option in plotly.offline.plot) resolves this.

It would still be interesting to know where plots are generated into.

[Mark Liversedge]

The directory you run GC in must be writable.

Thanks for the update: I have added "sip" to the modules that must be installed on the wiki page.
https://github.com/GoldenCheetah/GoldenCheetah/wiki/UG_Special-Topics_Working-with-Python

Mark

[Mark Liversedge]

Thanks Mick, its being looked at.

[Chris C.]

I see there was a patch for Hi-DPI, it's still a problem on my mac (Cinema HD 2560x1600)

[Manuel Oberti]

With the new release I can not add the 3d charts on activity page.

[Mark Liversedge]

You can adjust the font settings in preferences, what version of MacOS are you running?
FWIW, 2560x1600 isn't really hi-dpi.

Mark

[Mark Liversedge]

That's right, it has been deprecated, as outlined in the release notes.

Mark

[Chris C.]

That fixed it, I changed font to Helvetica Neue (from Helvetica) and moved the font size slider (and placed back at the same spot) and the top menu sizing has now bee resolved.

I also thought it's wasn't a large resolution but that's what it was referred as last time when I reported it.

C.

[Mark Liversedge]

Cool. thanks for feedback.

[Manuel Oberti]

Hi Mark,

the link to the video "todo tutorial" in UG_Special Topics_Working with Python does not work

Il giorno venerdì 16 febbraio 2018 09:44:04 UTC+1, Mark Liversedge ha scritto:

[El Duderino]

Hello,

since I´ve installed 1802 I got an error message as soon I start GC still bevor the "percentage window" of GC pops up. DEV 1710 work fine.

Windows 7

R installed

Other Laptop with Windows 10 works fine.

Greets

Am Freitag, 16. Februar 2018 09:44:04 UTC+1 schrieb Mark Liversedge:

[Ale Martinez]

El martes, 20 de febrero de 2018, 4:13:14 (UTC-3), Manuel Oberti escribió:

Hi Mark,

the link to the video "todo tutorial" in UG_Special Topics_Working with Python does not work

TODO: something to be done, the Python documentation is work in progress.

 

[Manuel Oberti]

Thanks Mark for your answers......but I have a couple of problem regardin python :

1- every time I close and reopen GC the flag "ENABLE PYTHON" are not selected. I have to flag the box, save and reopen

2- I have installed all the modules in the python wiki , but when I copy and parte your code 

import numpy as np
import plotly
from plotly.graph_objs import Scatter, Layout

## Retrieve power and cadence
xx = np.asarray(GC.series(GC.SERIES_WATTS))
yy = np.asarray(GC.series(GC.SERIES_CAD))

## Prepare Plot
f =plotly.offline.plot({
    "data": [Scatter(x=xx, y=yy, mode = 'markers')],
    "layout": Layout(title="Power / Cadence")
}, auto_open = False)

## Load Plot
GC.webpage(f)

This is the error.

GC folder  are writable , all the modules are installed , set path python 3.6 

The file plot-temp.html is created but I don't see any charts

THANK MARK

Il giorno venerdì 16 febbraio 2018 09:44:04 UTC+1, Mark Liversedge ha scritto:

[Manuel Oberti]

Hi mark....found the solution :

If I add in the code " filename="C:/Temp/power-cadence.html" the chart is shown correctly

Probably the GC folder is not configured for writing

THANKS

[Erik Botö]

I tried installing on a MacBook Pro running High Sierra, but I can't get Strava sync working.

I go through wizard of adding another cloud account, everything seems to be going well but when I later try to sync there's an error about not having a token setup.

Ideas?

Cheers,

Erik

[Ward Muylaert]

Filtering of activities seems to be completely broken on my end (macOS high sierra 10.13.3). Trying something like

Sport="Run"

returns all activities where Sport is "Bike". Using the same but with "Bike", returns nothing. Doing so with "Swim" returns Runs and Bike rides. Similar things happen with other text fields. (I think it might depend on the order they appear somewhere? For Sport the settings say "Bike,Run,Swim,Walk")

Trying "Duration < 3600" returns only activities with a duration longer than an hour. "Duration > 3600" returns all activities. Similar things happen with other numeric fields.

I see some changes for Python and R to the DataFilter files. I have both Python and R disabled though. Is it possible parsing was broken with the changes to DataFilter.l?

[Mark Liversedge]

Did you get thru the entire setup?
Did the Finish button?
Beware of dialogs getting hidden by the MacOS window manager.

[Mark Liversedge]

Make sure the search/filter box is in filter not text search mode.

[Erik Botö]

That was it!

After login in to Strava I got a msgbox about it being successful, but the wizard wasn't done and was hiding in the background.

Thanks,

Erik

[Portaner]

I have update "GoldenCheetah_V3.4_QT551_64bit.dmg" to "GoldenCheetah_35DEV1802_QT594_64bit.dmg" because I want to use HRV function.

Now my running training from 13.02.2018 has a value of 4482. In version V3.44 the value of 55 seems to me to be more realistic.

[wooliethai]

My trailing 30-day mileage formula has stopped working in this version

sum(Distance[[Date-30:Date]])*0.621371*7/30

What is most likely the issue?

[Ward Muylaert]

It is in filter mode, not text search. My existing custom filters are also broken because of this issue.

[Ward Muylaert]

I too have this issue.

[Mark Liversedge]

I cannot recreate this. Can you close GC and delete your rideDB.json file in the athlete cache directory and then restart GC please.

Mark

[John Gianni]

Just tried to clear cache, to no avail. Filters seems to be broken. 

IE: Device = "Zwift" returns all the activities recorded with Garmin :) ... Device <> "Zwift" returns all activities !!! Data contains "P" returns nothing even if I have power data everywhere. 

[Ward Muylaert]

Deleting rideDB.json did not fix it for me either.

[Mark Liversedge]

OK. Will check on Mac build later this week. I can't recreate at all.

[Mark Liversedge]

John - can you confirm you are on Mac too.

Mark

On Monday, 26 February 2018 10:53:56 UTC, John Gianni wrote:

[Mark Liversedge]

Ward,

Would it be possible to post a screenshot when a filter is active with the ride list visible please?
You can send it to liver...@gmail.com if you'd rather not post it publicly.

THANKS

Mark

[John Gianni]

Thanks, I confirm I'm on Mac.

[Ward Muylaert]

[Mark Liversedge]

Thanks !

It looks like a real bug, but totally weird that it only manifests itself on Mac. Will take a peek when near my Mac on Friday.

Mark

[Erik Botö]

I just tried building myself using Qt 5.10.1 and Qt 5.9.4 on MacOS and then I don't see the filter bug. Starting the 1802 dev release I see the same bug as multiple people reported.

Will be fun to see what's going on here.

Cheers,

Erik

[Arturo Hernández Rodríguez]

Hola Mark, no hay version para Linux?

[Mark Liversedge]

Erik -- you see this bug with 5.9.4 but not 5.10 ?

[Erik Botö]

No, I started out building with Qt 5.10.1 and didn't see the bug. So I figured I wanted to test 5.9.4 as well since that's what the release use but don't see the bug there either. Then I just wanted to mention both tests just to provide as much info as possible, but I guess I just made it unclear instead. :-)

But if I start the 1802 release I do see the bug, and that's using the exact same athlete and meta data.

Cheers,

Erik

[Mark Liversedge]

That is weird !

I'll test and do a full rebuild and then repost the binary.

Mark

[Mark Liversedge]

Gareth is working on it.

[Manuel Oberti]

Why am I in this situation with python turned off whenever I start GC?

[Mark Liversedge]

I can't read that screenshot it is too low res.

[Manuel Oberti]

Sorry Mark.......

I'll try in this way

THANKS

[Mark Liversedge]

Hi Erik,

Are building with python support?

FWIW I did a clean build on MacOS and getting the same bug as reported here. Really weird.

Mark

On Tuesday, 27 February 2018 11:34:44 UTC, Erik Botö wrote:

[Erik Botö]

No I did a very basic build, vanilla gcconfig.pri.in with the following additions

DEFINES += NOWEBKIT

LIBZ_LIBS    = -lz

I'm using Qt installed via the online installer, so I haven't built Qt myself. Xcode version is 9.2.

If you have anything you want me to test I'd be happy to help out.

Cheers,

Erik

[Ale Martinez]

May be the bug is related to the new scripting facility which is included only when Python/R support is enabled?

[Mark Liversedge]

I'm trying to isolate it, but this only manifests itself on MacOS which makes me wary of toolchain issues too.

Mark

[Mark Liversedge]

OK, it was a local build issue. I had a really old and out of date build in src/ that mean't flex/bison files in the build directory were not being used.

Will post a fixed MacOS binary later this evening after some testing.

Mark

[Argo]

Hello Mark,

I recently discovered GC for analyzing my running data, and I'm really impressed by the capabilities of the tool.  I'm currently using GC 3.4
Some days ago I updated from GC 3.4 to GC 3.5 dev 1802 on Windows 10 64bit, but the new dev version does not start on my system - or start but crashes very soon.
When I start it I see the windows hourglass for a short moment, but no windows appears. I've unistalled and reinstalled the c++ redistributeable several times, but it did not help.
I've searched also fro a crash log file but I could not find one.
According to the windows event viewer there seems to be a problem with ucrtbase.dll.
I've also tried the GC 3.5 dev 1710 build but with the same result.


Do you have any hints for me what to check or how to fix the problem?

Thanks in advance

Best Regards
Argo

[Zim Bou]

I'm having trouble with pointing GC to python, I have quite some versions installed but the version used is in /anaconda/bin/python. If I point GC to this folder an error message pops up saying "python does not appear to be installed in that location". After pointing GC to another python location in the path python actually crashes. The same happens when I start GC with that path specified.

There are quite some versions of python in my path (see screenshot), could that be the problem?

It's on a mac with GC version DEV-V3.5

Thanks!

[Mark Liversedge]

You need to point to python home, not the binary.
Be careful with anaconda as it comes with an old version of sip that is not compatible with GC 3.5

Mark

[Zim Bou]

Thanks! Unfortunately I still don't manage, I've set the path to /Library/Frameworks/Python.framework/Versions/3.4 as version 3.6 is the version installed with anaconda. However after that GC crashes on startup and could only be started again using "open ./GoldenCheetah.app --args --no-python". /Library/Frameworks/Python.framework/Versions/3.6 doesn't work either...

Unfortunately I can write python scripts but struggle with these system issues, even when in /Library/Frameworks/Python.framework/Versions/3.4 the command python still starts v.3.6.6 (the anaconda version). As to sip, there are currently two versions installed it seems:

sip                       4.18                     py36_0 
sip                       4.19.8                    <pip>

I've tried setting the path to /anaconda which contains python.app but that doesn't do the trick either...

Op vrijdag 9 maart 2018 19:45:58 UTC+1 schreef Mark Liversedge:

[Mark Liversedge]

sip must be 4.19.

PYTHONHOME is nothing special in GC, it is exactly as it should be set for anything else.

Mark

[Zim Bou]

Thanks again! It's working now, pointed GC to /Library/Frameworks/Python.framework/Versions/3.6 instead of /Library/Frameworks/Python.framework/Versions/3.6/bin and also cleaned up $PATH, maybe that did the trick. Will install sip 4.19 using pip, somehow conda install sticks with 4.18

Op zaterdag 10 maart 2018 10:53:04 UTC+1 schreef Mark Liversedge:

[Zim Bou]

I should add, I'm very excited about this and have been looking forward to using python in GC for quite a while so many thanks for implementing this!

Op zaterdag 10 maart 2018 18:32:17 UTC+1 schreef Zim Bou:

[Argo]

Am I really the only one not being able to get GC 3.5 running on windows.
Would be really gratefuly for any help or hint.

[Jörn R.]

I checked the .dll version on my system - it's exactly the same - we have about 500 Download for 64bit - and not heard of that. When searching for the general

error there were multiple of such reports - but no clear indication on the root cause. One hint occuring more the once pointed to "outdated" drivers - video,...

not really a clear direction/root cause indication - but if you please check your drivers/update where need maybe it helps. Sorry - so far no further idea. 

[Argo]

Jörn, thanks for your answer. I just tried to update my video driver, but it was already up to date.
I will go on searching for the reason of the crash.

[Argo]

Strike!! GC 3.5 is running.
The problem was, that on my Widows system an older version of Python (3.5.1) was installed. Obviously GC did not like that. After uninstalling Python, GC works.

[Mark Liversedge]

It should work with older versions of python 3.

Can you post the output of
> pip3 freeze

Or use pip if you installed v3 only

Mark

[Argo]

Hello Mark,
I think its too late to run pip, as I've uninstalled python.
Regards
Argo

[Argo]

Maybe the reason for my Python problem was that I had a 32bit and a 64bit version of Python installed in parallel.
Anyway, I'm happy that GC 3.5 is running now.

[Stefan]

Running GC on three win computers. On two I got Python running (though I ran in all sorts of "challenges", not as easy as getting R set up, definitely more/too involved. No criticism, just [dumb] user feedback).

However, on the third computer I'm stuck. Windows 10. Something with numpy. I've reinstalled numpy with pip but this didn't help. I'm actually not doing anything differently than on my other two computers. Searching for the error message did not really yield anything helpful for me. Has anyone encountered this ?

[Mark Liversedge]

you're running 32 bit python with 64 bit GC

[Matthew B.]

3.5-DEV1802 is consistently crashing for me during the sync / download phase from Strava on my Mac running MacOS 10.13.3.

Attached is the crash report. Let me know what else I can provide to help diagnose this problem.

Cheers,

Matthew

[Matthew B.]

Mark,

FYI: I downloaded GoldenCheetah_35DEV1802_QT594_64bit_MacOS_Fixed.dmg today, and the crash report is against this version of GoldenCheetah.

[Mark Liversedge]

The issue is regarding PD model estimates, the import whilst PDEstimates are being calculated is the main issue.

[Miloš Žarković]

I have installed VERSION 3.5 DEVELOPMENT BUILD 1802, Windows 10 and R 3.4.3 64bit
I got a strane problem in R chart:
If I plot data using plot(d$time,d$heart.rate) everything is OK
if I use plot(d$heart.rate~d$time) I got error:

Error in inDL(x, as.logical(local), as.logical(now), ...) :
unable to load shared object TC:/Programs/Sci/R/R-3.4.3/library/stats/libs/x64/stats.dllr:
LoadLibrary failure: The specified module could not be found.
In addition: There were 18 warnings (use warnings() to see them)

if I use hist(d$heart.rate) I got the same error.

R works perfectly in Rstudio and in plain installation

Thanks,
Miloš

[Argo]

I had this problem as well. I don't know the 'official' way to solve it, but in my case it helpes to add the folder of R x64 to the windows PATH.
I your case add C:/Programs/Sci/R/R-3.4.3/bin/x64 to the windows PATH variable. Afterwards restart GC.

Good luck.

[Miloš Žarković]

Thank you. R is working
Miloš

[Mark Liversedge]

Yes, the DLL could not be found.

[Hwa Runner]

Hello Mark,
I'm using since a few days GC 3.5 dev 1802 and I really like it - especially the new overview tiles, which look really cool.
Nevertheless I want to report some minor problems which I found.

1. The Trends Summary page does not correctly store the configured date range of the chart settings. I case I define that the date range to consider should be 'This Month', GC will show 'This Week' after restart.
See screenshot chartSettings.jpg

2. This one is no bug but just a wish. In the activities performance chart view, the scale on the y axis uses integer values. This is sometimes a bit confusing for data where only the position after the decimal point changes. Maybe ist is possible to show one digit after the decimal point in certain cases.
see screen shot ActivityCharts.jpg

3. Font size of Trends Summary in the Sidebar. On my High DPI Screen the fonts are that small compared to the fonts of the other panels or the menu, that I hardly can read them. Maybe it is possible to make them a little bit bigger.

See screenshot  summaryFont.jpg

I did my test on the 64bit Windows version.

Regards
Hwa

Am Freitag, 16. Februar 2018 09:44:04 UTC+1 schrieb Mark Liversedge:

[Nigel Van de Velde]

Any chance for a linux build this time?

Kind regards,

Nigel

[TommyRaton]

Are you looking for a *1802.deb?  Only because the git is available for a linux build with the most bleeding edge GoldenCheetah.  I'm darn close to totally incompetent but was able to get it working (libusb, vlc, kml, etc) on ArchLinux (except I can't get Python to make)

[Mark Liversedge]

There is some work being done to make a snap for GC, this should help in the future, see here:
https://github.com/GoldenCheetah/GoldenCheetah/issues/2808

Mark

[Portaner]

Hi,

I have look on my windows PC the problem with the file date from 13.02.2018 and there is the same problem we on my MacBook (see my Post from 23.02.2018 in this thread).

Also I have also found on date 27.01.2016 the same big GOVSS different between the GC version. On GC V3.4 the GOVSS is 142 and on GC V3.5 is 1170.

Maybe someone can pose this problem on their computer. Put my two training files with the CV available.

[Ale Martinez]

Both files have anomalous speed values at the start, use the Anomalies button in Edit tab to find and fix them and save so the GOVSS value is updated.

In the 13.02.2018 activity is just the first sample, the other the first 3 are wrong.

[Portaner]

Thanks Ale, now the GOVSS are right.

What I do not understand is, that GC V3.4 has apparently correctly calculated GOVSS with these "erroneous values".

I will take it that way.

[Ale Martinez]

El lunes, 19 de marzo de 2018, 14:00:56 (UTC-3), Portaner escribió:

Thanks Ale, now the GOVSS are right.

What I do not understand is, that GC V3.4 has apparently correctly calculated GOVSS with these "erroneous values".

Me neither...