VERSION 3.6 DEVELOPMENT BUILD

[Mark Liversedge]

https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-DEV2005

First development build in 3.6, no planning related functionality yet, but quite a lot of other stuff.

As ever report experiences and  issues here, and take a backup before you install !

Mark

[José Luis García Pallero]

First of all, thank you very much for your effort in this fantastic
software. I've installed the new development (I'm using Debian Sid)
and I would like to make some comments:

About the appearance:

1. I see the toolbar with Trends, Activity, etc. options was moved to
a left pane on the layout. I find this a bit impractical as this
situation "eats" space for the different analysis tabs, so they must
be selected using displacement. I think the original position (until
3.5 version) was more practical, as it was located in an area with any
information that now is blank. Attached I send an explained picture

2. About the analysis tabs (Summary, Trackes, Stress, etc.) I find in
theis version they are (the titles) more separated than in previous
versions, so in the same width are fewer tabs. Could be the width of
the tab title configured or at least fit to the title length? On the
other hand I find marvelous the new feature of move and change the
tab's order. Great!

About Python:

I can see Python 3.7 is the default Python version. It is great as 3.7
is the default version in most Linux distributions. But my Python
charts do not work anymore. In the attached figure the behavior can be
seen. The Python code does not inform about any error nor warning, but
the plot does not appear

Thanks

El dom., 3 may. 2020 a las 16:54, Mark Liversedge
(<liver...@gmail.com>) escribió:

> --
> _______________________________________________
> 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/332b6e08-aa7b-43ca-bedc-0369c6fffbff%40googlegroups.com.



--
*****************************************
José Luis García Pallero
jgpa...@gmail.com
(o<
/ / \
V_/_
Use Debian GNU/Linux and enjoy!
*****************************************

[Mark Liversedge]

On Sunday, 3 May 2020 20:55:40 UTC+1, José Luis García Pallero wrote:

About the appearance:

1. I see the toolbar with Trends, Activity, etc. options was moved to
a left pane on the layout. I find this a bit impractical as this
situation "eats" space for the different analysis tabs, so they must
be selected using displacement. I think the original position (until
3.5 version) was more practical, as it was located in an area with any
information that now is blank. Attached I send an explained picture

We are transitioning to a new UI.

See the slides 10-13 here: https://docs.google.com/presentation/d/1h8o6GbJJUhe3jq63rcyv7nVd01u8ETIcJZKtzqbDPXc/edit?usp=sharing

2. About the analysis tabs (Summary, Trackes, Stress, etc.) I find in
theis version they are (the titles) more separated than in previous
versions, so in the same width are fewer tabs. Could be the width of
the tab title configured or at least fit to the title length?

No.

About Python:

I can see Python 3.7 is the default Python version. It is great as 3.7
is the default version in most Linux distributions. But my Python
charts do not work anymore. In the attached figure the behavior can be
seen. The Python code does not inform about any error nor warning, but
the plot does not appear

Make sure you 3.7 install has the required modules available, the console will show errors too, check them.

See: https://github.com/GoldenCheetah/GoldenCheetah/wiki/UG_Special-Topics_Working-with-Python#recommended-modules-and-use-of-pip

Mark

[TJDunnigan]

I don't see this in the release notes, but wonder if libusb 1.02 (or higher) was incorporated into the Mac build (since it is in previous development builds)?

 MacOS i

[Ale Martinez (Please don't email or cc me)]

El domingo, 3 de mayo de 2020, 19:14:27 (UTC-3), TJDunnigan escribió:

I don't see this in the release notes, but wonder if libusb 1.02 (or higher) was incorporated into the Mac build (since it is in previous development builds)? 

yes, libusb 1.0.23, also D2XX:

GoldenCheetah.app/Contents/MacOS/GoldenCheetah --version

GoldenCheetah V3.6-DEV2005 (4000)

Build date: May 3 2020 02:53:07

Build id: 4000

Version: V3.6-DEV2005

DB Schema: 154

Metrics: 361

OS: Mac OS X 10.13

QT: 5.14.2

QWT: 6.1.1

GCC: 4.2.1

SRMIO: yes

D2XX: yes

KML: none

ICAL: 3.0

USBXPRESS: none

LIBUSB: yes (1.0)

VLC: none

VIDEO: none

SAMPLERATE: 0.1.9 

SSL: yes

R: none [4.0.0]

Python: none [3.7.7]

WEBKIT: none

LMFIT: 7.0

LEVMAR: 2.6 (November 2011)

[José Luis García Pallero]

El dom., 3 may. 2020 a las 22:18, Mark Liversedge
(<liver...@gmail.com>) escribió:
>

This is what I can't understand. The console does not show any error,

but the plot does not appear

Cheers

>
> Mark
>
> --
> _______________________________________________
> 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/5dbb6807-7702-4880-bd2b-662cf215275b%40googlegroups.com.

[pepe]

Hi, many thanks for 3.6 DEV2005

I see three small glitches

1   Installing in windows 8.1 64bit: everything apparently good, no install errors. But when launched GC, it  crashed with an error indicating that Qt5Bluetooth.dll version only works with win10.   Workaround:   Copied Qt5Bluetooth.dll from version 3.5, and it seems to work.  Test trying to find Polar H10 and Stryd without luck at BT4 discovery.

2.  When adding a Withings account it is not possible to type symbol "@" in the email entry box, workaround typed my email in notepad and copied pasted.

3. In training view, error indicating that libwinpthread-1.dll and libgcc_s_seh-1.dll where missing

Thanks

[Mark Liversedge]

On Monday, 4 May 2020 05:32:58 UTC+1, pepe wrote:

Hi, many thanks for 3.6 DEV2005

I see three small glitches

1   Installing in windows 8.1 64bit: everything apparently good, no install errors. But when launched GC, it  crashed with an error indicating that Qt5Bluetooth.dll version only works with win10.   Workaround:   Copied Qt5Bluetooth.dll from version 3.5, and it seems to work.  Test trying to find Polar H10 and Stryd without luck at BT4 discovery.

Thanks - bad news I'm afraid, Bluetooth is only supported on windows 10 via UWP (Universal Windows Platform).

Mark

[TJDunnigan]

Thanks, Ale

[Valvero]

Does the new version not have the Diary tab?

[Mark Liversedge]

On Monday, 4 May 2020 20:30:31 UTC+1, Valvero wrote:

Does the new version not have the Diary tab?

No. It is being deprecated in 3.6 and replaced with a planning mode.

Mark

[Valvero]

Planning mode is not yet available, right?

Is there any possibility of adding the diary?

[Ale Martinez (Please don't email or cc me)]

El lunes, 4 de mayo de 2020, 17:02:05 (UTC-3), Valvero escribió:

Planning mode is not yet available, right?

Is there any possibility of adding the diary?

Not the previous diary view, but if you just want to recover the full calendar chart, importing the attached file will recreate it on Activities view with your previous settings.

[Martin Wifling]

If you have exported the charts in diary view before the update, it should be possible to re-import them. When selecting at the sandwich menu the IMPORT DIAGRAM option, you can now choose to import in in trends view.

[Valvero]

Perfect. I have already imported it. Thank you

[UlysseUlysse]

Merci ! Super !!

Mark

[José Luis García Pallero]

El dom., 3 may. 2020 a las 16:54, Mark Liversedge
(<liver...@gmail.com>) escribió:
>

> https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-DEV2005
>
> First development build in 3.6, no planning related functionality yet, but quite a lot of other stuff.
> As ever report experiences and issues here, and take a backup before you install !

Hello:

I'm having troubles with Python graphics and laps. This piece of code:

laps = GC.activityIntervals()
pos = -1
for i in range(len(laps["selected"])):
if laps["selected"][i]:
pos = i
break
print(i)

shows on screen the internal number of the selected lap in an
activity. The problem is that now, with version 3.6, when I select an
interval the code is not executed again, so it is impossible to change
the interval and refresh the plot (if another tab is selected and then
I back again to the Python graphic it is refreshed). In version 3.5
the plot was refreshed without problem

>
> Mark
>
> --
> _______________________________________________
> 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/332b6e08-aa7b-43ca-bedc-0369c6fffbff%40googlegroups.com.

[Alexander Voos]

Got problems with python carts now.

Did the Updates for Python and R and SIP but Python-Charts don't work.

They had worked before.

May someone can help.

Kind Regard 

Alex

[Mark Liversedge]

On Tuesday, 5 May 2020 12:31:07 UTC+1, Alexander Voos wrote:

Got problems with python carts now.

Did the Updates for Python and R and SIP but Python-Charts don't work.

They had worked before.

You need to install the dependencies, esp. SIP, see the instructions here:

https://github.com/GoldenCheetah/GoldenCheetah/wiki/UG_Special-Topics_Working-with-Python#recommended-modules-and-use-of-pip

Mark

[Mark Liversedge]

On Tuesday, 5 May 2020 11:24:42 UTC+1, José Luis García Pallero wrote:

The problem is that now, with version 3.6, when I select an
interval the code is not executed again

This is a regression, will look to fix it for the next development build.

Mark

[José Luis García Pallero]

Hello:

In this new v3.6, as in the previous versions, I can see in the
Athlete->Measurements that when introduced manually, there are two
related fields which must be filled individually: Fat percent and Fat
mass. As one can be computed based on the other I propose to set these
fields as mutually exclusive, so if the user introduces the fat
percentage the fat mass is automatically computed, and vice-versa.
Actually the fields do not need to be programmed as mutually
exclusive. For example, if the user fills the fat percentage and
leaves fat mass as blank or zero, this last field must be computed by
GC

El dom., 3 may. 2020 a las 16:54, Mark Liversedge
(<liver...@gmail.com>) escribió:
>

[RB]

Hi,

First of all thx again for the great effort you guys are putting into this development

I got some issue using the new GC3.6

1. My calculated values like CP,  W' and P-Max on an activity are empty (displays NA on summary of an activity)

2. In the trends view e.g. Library chart "CP History" CP estimated values are not displayed as well

What i tried so far is removing files in the cache directory and reset layout, on trends and activity view.

Any suggestions (is this due the upgrade to 3.6 or something else)?

Kind Regards,

Remco

[Mark Liversedge]

On Saturday, 9 May 2020 08:43:53 UTC+1, RB wrote:

I got some issue using the new GC3.6

1. My calculated values like CP,  W' and P-Max on an activity are empty (displays NA on summary of an activity)

2. In the trends view e.g. Library chart "CP History" CP estimated values are not displayed as well

There are not changes that I know of that should cause this, it may be a coincidence.

It sounds like estimates had not been generated- possibly because a metric refresh was in process?

Mark

[RB]

Hi Mark,

Viewing the change log I also did not expected this.

In the case of an coincidence, I suspected that when i deleted the rideDB.json and start GC again wait till GC is finished with creating an new rideDB.json start GC again, that the problem should be gone.

unfortunately.

I now also tried creating a new user reimporting activities via activity json files and also tried via new user and importing from source fit files. Also for these new users have no model (W' CP P-max) figures are calculated.

To be clear metrics like CTL,ATL, TSB etc are calculated correctly.

What am i missing any suggestion on how to solve/analyse this further.

Kind Regards,

Remco

[Mark Liversedge]

On Saturday, 9 May 2020 10:22:38 UTC+1, RB wrote:

What am i missing any suggestion on how to solve/analyse this further.

Its a bug- I just checked and I have the same problem- its not a build issue, its a bug.

Will look into it.

Mark

[Ale Martinez (Please don't email or cc me)]

El sábado, 9 de mayo de 2020, 4:43:53 (UTC-3), RB escribió:

Hi,

First of all thx again for the great effort you guys are putting into this development

I got some issue using the new GC3.6

1. My calculated values like CP,  W' and P-Max on an activity are empty (displays NA on summary of an activity)

2. In the trends view e.g. Library chart "CP History" CP estimated values are not displayed as well

For the case 2. a workaround this regression is to re-add the curve to the chart, there is no workaround for 1. AFAIK

[Mark Liversedge]

On Saturday, 9 May 2020 16:36:05 UTC+1, Ale Martinez (Please don't email or cc me) wrote:

For the case 2. a workaround this regression is to re-add the curve to the chart, there is no workaround for 1. AFAIK

Thanks Ale,

The rename needs to stay, so here attached a chart to drop in and replace the existing one.

Cheers

Mark

[UlysseUlysse]

hello Mark
Problem with GoldenCheetah_v3.6-DEV2005_x64.AppImage

The average table for "Trends" and "Activities" no longer displays the results

[Mark Liversedge]

On Sunday, 10 May 2020 14:17:33 UTC+1, UlysseUlysse wrote:

hello Mark
Problem with GoldenCheetah_v3.6-DEV2005_x64.AppImage

The average table for "Trends" and "Activities" no longer displays the results

Thanks, this has been reported a few times now, and fixed by Ale.

I need to fixup the Datafilter, but suspect very few people noticed that one :)

https://github.com/GoldenCheetah/GoldenCheetah/commit/8627f26b2c8e6f8c9fcd364043fb746512914760

Mark

[Jozef]

Good evening, thank you for your super job. I installed dev 3.6 on my windows 10 pc. Previously I had version 3.5. However, the program crashes. The following message is reported to me:
Unhandled exception 0xc0000005
RAX = 2 RCX = 1c7b7c24b58 RDX = ffffffff RBX = 2
RSP = 2b7efc180 RBP = 1 RSI = ffffffff RDI = 1c7b7c24b58
R8 = 2b7efc1b0 R9 = 2b7efc1b0 R10 = 3 R11 = 246
R12 = 0 R13 = 0 R14 = 7ffa7f357760 R15 = 7ffa7f623e88

Qt5Core.dll! QBasicMutex :: lockInternal + 0xc0
Qt5Network.dll! QNetworkConfigurationManagerPrivate :: configurationAdded + 0xd3
Qt5Network.dll! QNetworkConfigurationManagerPrivate :: qt_static_metacall + 0xfd
Qt5Core.dll! QObject :: event + 0x183
Qt5Core.dll! QCoreApplication :: notifyInternal2 + 0xA4
Qt5Core.dll! QCoreApplicationPrivate :: sendPostedEvents + 0x22b
Qt5Core.dll! QEventDispatcherWin32 :: processEvents + 0x70
Qt5Core.dll! QEventLoop :: exec + 0x1eb
Qt5Core.dll! QThread :: exec + 0x89
Qt5Core.dll! QThread :: start + 0x36d
KERNEL32.DLL! BaseThreadInitThunk + 0x14
ntdll.dll! RtlUserThreadStart + 0x21

How can I solve it?

[Ale Martinez (Please don't email or cc me)]

El martes, 12 de mayo de 2020, 14:04:09 (UTC-3), Jozef escribió:

Good evening, thank you for your super job. I installed dev 3.6 on my windows 10 pc. Previously I had version 3.5. However, the program crashes. The following message is reported to me:
Unhandled exception 0xc0000005
RAX = 2 RCX = 1c7b7c24b58 RDX = ffffffff RBX = 2
RSP = 2b7efc180 RBP = 1 RSI = ffffffff RDI = 1c7b7c24b58
R8 = 2b7efc1b0 R9 = 2b7efc1b0 R10 = 3 R11 = 246
R12 = 0 R13 = 0 R14 = 7ffa7f357760 R15 = 7ffa7f623e88

Qt5Core.dll! QBasicMutex :: lockInternal + 0xc0
Qt5Network.dll! QNetworkConfigurationManagerPrivate :: configurationAdded + 0xd3
Qt5Network.dll! QNetworkConfigurationManagerPrivate :: qt_static_metacall + 0xfd
Qt5Core.dll! QObject :: event + 0x183
Qt5Core.dll! QCoreApplication :: notifyInternal2 + 0xA4
Qt5Core.dll! QCoreApplicationPrivate :: sendPostedEvents + 0x22b
Qt5Core.dll! QEventDispatcherWin32 :: processEvents + 0x70
Qt5Core.dll! QEventLoop :: exec + 0x1eb
Qt5Core.dll! QThread :: exec + 0x89
Qt5Core.dll! QThread :: start + 0x36d
KERNEL32.DLL! BaseThreadInitThunk + 0x14
ntdll.dll! RtlUserThreadStart + 0x21

How can I solve it?

1) do not install over the top of an existing install - use a different folder or uninstall first

2) shutdown and restart after installing the VC C++ redistributable (like the dialog says).

There is another thread on this issue: https://groups.google.com/forum/#!topic/golden-cheetah-users/95eXJXeHq68

[Eric Christoffersen]

On windows 10 I'm getting much more frequent hangs when terminating a virtual ride session. Maybe 70% of the time now it is hanging on vlc shutdown. I was hoping this was going to be fixed by magic when upgrading to the new qt.

I believe this is related to the overlay widgets as the hangs started for me when the widgets first appeared.

Anyone else seeing this? I can recover the train file from the logging but is still pretty frustrating.

Wondering if anyone knows exactly how widgets and vlc interact. Perhaps there needs to be a synchronous notification to the widgets tell them to stop before we try to close vlc?

Thanks,

Eric

[Ale Martinez (Please don't email or cc me)]

El martes, 12 de mayo de 2020, 14:59:48 (UTC-3), Eric Christoffersen escribió:

On windows 10 I'm getting much more frequent hangs when terminating a virtual ride session. Maybe 70% of the time now it is hanging on vlc shutdown. I was hoping this was going to be fixed by magic when upgrading to the new qt.

Our hope was to replace vlc with qt5 video for 3.6 to remove dependency on older vlc 2.2, but there are problems with videosync, from Jörn tests:

- For "normal" Video replay QT5 video functionality (at least on Windows) works fine and works without a noticeable difference to VLC

- In RLV mode QT5 Video is not working as have hoped and wished. Main problem is that when adjusting the replay speed (which is possible - similar to VLC) there is visible "delay" in form of a black screen before the new speed applies. In VLC this happens totally smooth without any breaks or visible artefacts when changing. This problem gets worse when changing replay speed every 1-2 seconds. But even if the change happens only every 10-20 seconds its visible and lasts partly 1-2 seconds before the new speed applies.

 

I believe this is related to the overlay widgets as the hangs started for me when the widgets first appeared.

If you want to test, truncating video-layout.xml to an empty file disables the widgets on the next start.

 

Anyone else seeing this? I can recover the train file from the logging but is still pretty frustrating.

Wondering if anyone knows exactly how widgets and vlc interact. Perhaps there needs to be a synchronous notification to the widgets tell them to stop before we try to close vlc?

I don't know, the current scheme doesn't even work on the other platform where we use VLC (Linux).

[Jozef]

Thanks for the reply. I had already tried to uninstall version 3.5, delete any trace and install 3.6 dev but the problem remains. Strange because I installed 3.6 regularly on a fixed PC and I didn't have any problems.

[Ale Martinez (Please don't email or cc me)]

El miércoles, 13 de mayo de 2020, 3:52:08 (UTC-3), Jozef escribió:

Thanks for the reply. I had already tried to uninstall version 3.5, delete any trace and install 3.6 dev but the problem remains.

The second step is relevant: reboot after installing the included the VS C++ redistributable

 

Strange because I installed 3.6 regularly on a fixed PC and I didn't have any problems.

May be a SFC and Windows update could help

[mondoshiwan]

Am Sonntag, 3. Mai 2020 16:54:48 UTC+2 schrieb Mark Liversedge:

https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-DEV2005

First development build in 3.6, no planning related functionality yet, but quite a lot of other stuff.

As ever report experiences and  issues here, and take a backup before you install !

Mark

3.6 dev build - the slopes aren't editable anymore?

would be no big deal with garmin FIT files, but i'm asking this cause i do have a lot of Ambit 2 SML files imported and still using it, and here the GC calculated slopes are more like randomly thrown in numbers.
For some years i had an external solution, computing the slopes out of the VerticalSpeed the original SML file holds. Meanwhile i was able to build a Python Fix, with the help of the temp example, using a rise over run math - both methods gave similar good results. This worked till Version 3.5

But…now, if an edited activity is opened with the 3.6 dev build all edits are gone - seems that GC is not using the stored values in the json file, but only numbers computed on the fly. And it's not possible to edit those slopes in dev 3.6, even with the python fix.

Opened in 3.5 again everything is fine.

examples
original imported sml file:

the fix:

btw:

did an python fix for altitude too, which fills up missing values in beginning and eleminates the barodrift over time. This is still working in dev 3.6!

and thx for all the effort you guys put in the development of GC. It's my favorite training/ analysing software!

Carsten

[Ale Martinez (Please don't email or cc me)]

El sábado, 30 de mayo de 2020, 17:25:24 (UTC-3), mondoshiwan escribió:

Am Sonntag, 3. Mai 2020 16:54:48 UTC+2 schrieb Mark Liversedge:

https://github.com/GoldenCheetah/GoldenCheetah/releases/tag/v3.6-DEV2005

First development build in 3.6, no planning related functionality yet, but quite a lot of other stuff.

As ever report experiences and  issues here, and take a backup before you install !

Mark

3.6 dev build - the slopes aren't editable anymore?

v3.6 includes a greatly improved Fix GPS which allow to smooth slope and route, but also a change to recompute slope as derived series even when it is already present, which is an error IMHO since it defeats the point of storing slope in JSON files, I think we should revert this part: https://github.com/GoldenCheetah/GoldenCheetah/commit/6e5fb4217194d4b896e7db680cf8e40f14d11c1c#diff-23f3a036915f75349203d34b3789ad0c

[Eric Christoffersen]

I agree. My main use for fix gps was to fix the altitude so the slope wouldnt be insane on a virtual ride. Never imagined that people would be want existing slope, let alone that it would be considered more correct. Im wondering, if adjusting slope why not fix altitude too? Then the derive wouldnt change the slope.

One side of my head objects to slope and altitude disagreeing, seems like a single series would suffice.

If derive change is reverted then fixgps will no longer change slope when it changes altitude. Should i add a check box to fixgps to give permission to overwrite slope? Or is that presumed when user asks to adjust altitude?

Id also like to see if the python thing works better than what fix gps is doing. Could use it instead of the bspline. Bspline works great as a general unbiased smoother but still not as good as being smart.

[Eric Christoffersen]

I just read the python, no magic there either. Oh well.

Carsten, in your case you will be a lot happier running the fixgps altitude smoothing pass automatically on import. That way no need to prepass and your altitude and slope will be in harmony. The root problem in your original data is the altitude.

That said it is certainly possible that someone recorded their route using a surveyor's inclinometer and has totally wonderful 6 digits of slope precision - we've no business overwriting that without asking.

[Eric Christoffersen]

Last thing, promise, the sml found another bug.

This time in the slope derivation. It sets slope to zero when 'run' is zero. We should duplicate the old slope when distance doesn't change.

[Ale Martinez (Please don't email or cc me)]

El domingo, 31 de mayo de 2020, 0:15:13 (UTC-3), Eric Christoffersen escribió:

I agree. My main use for fix gps was to fix the altitude so the slope wouldnt be insane on a virtual ride. Never imagined that people would be want existing slope, let alone that it would be considered more correct. Im wondering, if adjusting slope why not fix altitude too? Then the derive wouldnt change the slope.

One side of my head objects to slope and altitude disagreeing, seems like a single series would suffice.

If derive change is reverted then fixgps will no longer change slope when it changes altitude. Should i add a check box to fixgps to give permission to overwrite slope? Or is that presumed when user asks to adjust altitude?

IMO, whe altitude is adjusted at user request it’s ok to fix slope, that’s what Fix Elevation does: https://github.com/GoldenCheetah/GoldenCheetah/blob/2d3eed0a98513f4f0e88d05dc73a64ec59e951a5/src/FileIO/FixElevation.cpp#L247 

[mondoshiwan]

Eric, don't know if i'm happier ;-) checked that method in 3.6 - yes it works quick and easy, but IMO the *Python thing* over that timerange gives a bit more reliable results.

The example i posted is an up and down course - have to run down from ~33m to sealevel in two steep sections. The first ist about12% - second around 8 to 9% and i tell you, this hurts on the way back ;-) especially the steeper one. The Altitude fix within 3.6 gives slopes around 7.7 here for the steeper section.

Will test it with more files, but i think it would be always good to have more than one option for fixing this, cause there will be no perfect method.

[Eric Christoffersen]

That is awesome that you like the averaging more. Like you say there's no perfect here.

When you tried fixgps did you play with the parameters? I believe you should be able to get something very close to correct, at least within the noise. Mathematically it should be able to do everything that averaging can do.

I think the real hassle with it today is that you need to iterate to get the route you expect. Really needs a map and altitude graph in the ui so you can see the results before applying. I'd like to see a picture that showed what regions were being declared outliers, that is the key to a good fit.

[mondoshiwan]

No didn't play with that parameters - was too overwhelming at the first look :) but will dive into!, I did just the altitude correction.

The proplem here at my running place with GPS is always to have a good reception, but this area is very challenging, lot of buildings and now covered by trees too. So the barometric readings are giving more better results. Suunto SML Altitude are with no digits and first i was using an external program with a math computing the slopes from the vertical speed - this data here is way more accurate than the alti readings and is useful for second by second computation and gave similar good results as my *python thing*. If anybody want's to try this too within the GC source code - the fomular is : (6 * Suunto_VerticalSpeed *60) / (Suunto_Speed * 3.6)

[Eric Christoffersen]

Did you try the ‘fix elevation’? I mostly mountain bike so fix elevation is terrible for me, data is too course, but in your case the mapquest data looks crazy detailed. Might be thats the best approach for you.

[mondoshiwan]

Am Dienstag, 2. Juni 2020 03:08:36 UTC+2 schrieb Eric Christoffersen:

Did you try the ‘fix elevation’? I mostly mountain bike so fix elevation is terrible for me, data is too course, but in your case the mapquest data looks crazy detailed. Might be thats the best approach for you.

yupp - that's what i did - only the elevation fix and this was'nt as good as the python fix or the correction from vertical speed.

[José Luis García Pallero]

Hello:

I can observe in this version that after closing (clicking the window close button) the program takes a significative amount of time bafore the efective close. At least this amount of time is significantly greater than for the GC 3.5. I run GC in Debian Sid with Plasma desktop environment on a Lenovo Thinkpad W540 Intel Core i7-4800MQ, 2.70GHz. I have a total amount of data of about 700 files (almost 1000 hour of data) and the delay between the close button click and the efective close is about 9 seconds. A similar time is employed to open GC, but in this case it is similar to the time for GC 3.5. I suppose this new version performs some operations before the program closes.

On the other hand, I can observe that in this version the Python charts are faster than in GC 3.5

[Fe]

Hi, I tried "Python fixes" but the window is locked, it doesn't show complete.

The function cannot be used.

Golden Cheetah 3.5 or 3.6 Windows 10
Thanks

[Ale Martinez (Please don't email or cc me)]

Is your Python install working on GC Python charts?

An alternative is to try with snapshots builds, which includes Python and dependencies preinstalled: https://groups.google.com/d/msg/golden-cheetah-users/6t-FcU8K5i8/U_WpkWBsBQAJ

[Fe]

Hi, the problem was already present since the feature was introduced in the Golden Cheetah DEV 3.5 (sop Windows 10 64)

Today I use the build 3.6 ci.3035 (02/06/2020) .... I also installed previous versions of GC with "Library" folder and all working with Python.

I have also installed other snapshots builds before.

The "Python fix" window is cut, the bottom of the window is not available. (look at the picture)

I use "Python Chart" without problems, today with Python 3.7.7.

One thing:
To have all the packages needed, for the various chart available, I copied the "Lib" folder from the original Python 3.7.7 installation.
Python chart work.

If I put the original Python path in the GC options (as always) it always connects to the Python(GC) version, the link does not change.

Thanks

[Ale Martinez (Please don't email or cc me)]

El miércoles, 3 de junio de 2020, 12:13:31 (UTC-3), Fe escribió:

Hi, the problem was already present since the feature was introduced in the Golden Cheetah DEV 3.5 (sop Windows 10 64)

Today I use the build 3.6 ci.3035 (02/06/2020) .... I also installed previous versions of GC with "Library" folder and all working with Python.

I have also installed other snapshots builds before.

The "Python fix" window is cut, the bottom of the window is not available. (look at the picture)

I see, the problem is the window is too big for your screen... You can resize, move or just maximize it to use all the screen space available.

[Fe]

Windows Text size is currently 125% if changing to 100% is really too small.
The Pytohn fix window is completely visible but impossible to read and write.
Thanks anyway

[Ale Martinez (Please don't email or cc me)]

I reduced the minimum height of that dialog 35%, hope it is enough.

[Fe]

Hi, perfect now the "Python fix" window is full.
Golden Cheetah 3.6 Build ci.3045

New :
I can't find the "More ..." menu on the charts... (??)

Great Job
Thanks

[Mark Liversedge]

On Thursday, 4 June 2020 13:24:29 UTC+1, Fe wrote:

Hi, perfect now the "Python fix" window is full.
Golden Cheetah 3.6 Build ci.3045

New :
I can't find the "More ..." menu on the charts... (??)

Menus have moved to the tab.

Mark

[Fe]

Ok, now I understand, thanks

[Eric Christoffersen]

Wrt video deadlock on windows 10, now with vlc 3.0.10.

What im finding is that i see none of these hangs until my machine has been booted for a few weeks. The hangs are on vlc stop, in d3d.

Ive tried all sorts of locking, pause before stop, waiting for vlc notifications, etc and when the hangs are happening nothing i do makes any difference. The stop call hangs in d3d.

Once i reboot i can go another few weeks before i can provoke a stop hang. I think root cause is something with the driver and d3d, some handles leaked? Physical memory fragmented? Guessing...

Ive read the vlc interface and its very well organized, the problem i think is in the interaction with the dwm and the driver, with dxva thrown in for good measure.

I dont see anything amiss with windows resource tools, no crazy handle counts, nothing in Ram of video memory is amiss. I would like visibility into the driver.

Of course could be some other temp data that doesnt persist across a reboot. Driver or vlc caching? Dwm?

No other apps on system show issues but this is the only video i play. Standalone vlc works fine with same files.

Fwiw im running on lenovo p50 with intel and nvidia graphics hardware. Would like to hear if other people are getting hangs on training ‘stop’.

[mondoshiwan]

On a Mac in Darkmode the Text within the Filter/Search window is black on dark background and only changing to white if the text is highlighted:

[Tim Sinclair]

Hi

THANK YOU  - 3.6 installed so easily  - Absolutely Perfect :-) !

I've just installed GC 3.6, as instructed and imported my training backups - everything works first time - fantastic job GC!!!

I have just one tiny problem that I wondered if you could help me with. I've searched through as many of the forums and held info as I can, but can't find a solution.

It's simply that the new Overview Tab for trends and activities does not appear on the tab bar by default. I can only open it via the 'Add Chart' sub menu. Once opened I would like to be able to permanently add the 'Overview' to the tabs bar, alongside the rest of the tabs at the top of the chart display. I've tried dragging it onto the tab bar etc, just in case that works - but sadly I can still only open it in a sepereate floating window.

Any fix for this would be VERY welcome.

I'm sorry if it's been dealt with elsewhere and I haven't seen it.

Sincerest best wishes and thanks for everything you do for us :-)

[Ale Martinez (Please don't email or cc me)]

El miércoles, 17 de junio de 2020 a la(s) 13:11:04 UTC-3, Tim Sinclair escribió:

Hi

THANK YOU  - 3.6 installed so easily  - Absolutely Perfect :-) !

I've just installed GC 3.6, as instructed and imported my training backups - everything works first time - fantastic job GC!!!

I have just one tiny problem that I wondered if you could help me with. I've searched through as many of the forums and held info as I can, but can't find a solution.

It's simply that the new Overview Tab for trends and activities does not appear on the tab bar by default.

Default layouts has not been updated yet, and they apply only to newly created athletes or Reset Layout.

 

I can only open it via the 'Add Chart' sub menu. Once opened I would like to be able to permanently add the 'Overview' to the tabs bar, alongside the rest of the tabs at the top of the chart display. 

I've tried dragging it onto the tab bar etc, just in case that works - but sadly I can still only open it in a sepereate floating window.

It is the same as with any other chart you want to add: press Ok and it will be added at the end of the current view.

[Tim Sinclair]

Thanks for your speedy response :-) That should have been obvious to me really ;-) :-)