Strengthen tracker's maximum distance safeguard by ensuring non-measurable distances are dismissed

[nirvn]

Trying to debug this weird situation where tracks will have "spikes" even though the maximum distance safeguard is one. This is how it looks like:

I have no idea what these spikes are, but the only way I could hypothesize we'd end up there is if a/ the position vertex that creates the spike is somehow invalid, which would then b/ create an exception when calculating the distance. Under that scenario, we were setting the distance to 0.0 which defeated the max. distance safeguard.

The PR fixes that.

[qfield-fairy]

🍎 MacOS DMG universal builds

Download a MacOS DMG universal build of this PR for testing.
(Built from commit 836638c)

📱 Android builds

Download an Android arm64 build of this PR for testing.
(Built from commit 836638c)

Other Android architectures

• arm-neon
• x64
• x86
• all access arm64
• all access arm-neon
• all access x64
• all access x86

🪟 Windows builds

Download a Windows build of this PR for testing.
(Built from commit 836638c)

🐧 Linux AppImage builds

Download a Linux AppImage build of this PR for testing.
(Built from commit 836638c)

[Ipickedausername]

Can't wait to check that ! I used to delete these errors manually back in the office.

[nirvn]

@Ipickedausername , interesting, are you suggesting that you were using a max. distance safeguard and yet would see these random spikes?

[nirvn]

@Ipickedausername , I'm asking because I have not been able to replicate this misbehavior and I was wondering whether you had a clear understanding of the circumstances around this.

[Ipickedausername]

I tested on two phones :

• Galaxy S9 > Every map was covered with these spikes.
• Galaxy XCover 4 (or 5, not sure) > No problem.

I ended up thinking that my phone might not be working properly, or that the issue could be related to using the app in areas with poor network coverage.

Most of these “spikes” appeared when my phone screen was off while I was walking with the app open. Almost every time I unlocked my phone to check the map, the tracking session created a spike.

The spikes were not pointing in the same direction, although several of them were often going toward the same point. Radom point, but sometimes not that random (e. g. : the hotel I stayed the night, the place I parked my car).

[nirvn]

@Ipickedausername , were you using the maximin distance safeguard to begin with?

[kochon]

we are also affected by this issue. it seems to only happen on android.

i will install the android build on my employee's device and report back with more details

[Ipickedausername]

@Ipickedausername , were you using the maximin distance safeguard to begin with?

Hmm, I'm not sure now.
My last try was in June last year, will be back on it starting mid-April.

[kochon]

Unfortunately the issue is still present on my employee's device. He mentioned to me that it seems to happen when either resuming tracking or when the screen is locked. We switched back to tracking on points layers for now.

[Ipickedausername]

Hi,
Same for me.
And the app stops working when I leave it in background, resulting in the loss of the tracks.

[nirvn]

@Ipickedausername , @kochon , if you guys have a samsung phone, I've been told turning off Samsung's "precise positioning" helps greatly.

I only have the screenshot in French, desole ;)

[Ipickedausername]

Oh indeed !

Here is a comparison 1 hour with "precise positionning" on and 1 hour without.

Why is the max distance limitation not blocking this behavior ?

Thank you for the tip !

[nirvn]

@Ipickedausername , I would sure love to know that. I'm investigating this further.

[Ipickedausername]

@Ipickedausername , I would sure love to know that. I'm investigating this further.

Maybe the max distance is calculated using the last calculated value, without considering if the last calculated one was valid or not. If 2 locations are detected 10km away from the user, but near each other, the first one, above n meters is canceled, but the second one, near the previously canceled one is consideres valid ?

Well, at least there is a way to avoir this issue on the user's end. Thank you !