New impact of "GPS week number rollover" in Free version of OM (7.4.26)

Started by LaurentG, September 12, 2022, 05:40:15 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

LaurentG

Hi Orux,

I recently discovered that the date of recorded tracks is now wrong on my phone, due to the "GPS week number rollover" (delta of 1024 weeks between actual and recorded date)

But I'm 100% sure it was OK on this same phone in 2021 and before (I have in my oruxmapstracks.db some tracks recorded at this time, in which dates are OK).
Why this change between 2021 and now ? Impact of "GPS week number rollover" was in 2019 april, so it was already the case for long in 2021, and nevertheless, there were no impact on my phone at this time.

Two facts could explain this change : update of system software on my phone, or update of OruxMaps itself.
First option can be definitely given up : my phone (that is quite old) does not receive any update for long time, last update was in 2018 or 2019...
On the other hand, in 2021, the "free" version of OruxMaps was 7.4.23 or 7.4.24, and we are now in 7.4.26

I don't know what are exactly changes between these versions (they are not documented, change log says only "Bug correction")

So my question is for Orux/Jose, who is the only one who can know : Which "fixes" in 7.4.25 or 7.4.26 (or maybe in 7.4.24) could explain this unwilled evolution ? Could you publish a 7.4.27 that would "repair" it ?

Thanks in advance

PS : I tried to reinstall 7.4.23 over current 7.4.26 (I kept the old APK), but the system rejected the update... maybe because it is a downgrade ?

orux

Quote from: LaurentG on September 12, 2022, 05:40:15 PM
Hi Orux,

I recently discovered that the date of recorded tracks is now wrong on my phone, due to the "GPS week number rollover" (delta of 1024 weeks between actual and recorded date)

But I'm 100% sure it was OK on this same phone in 2021 and before (I have in my oruxmapstracks.db some tracks recorded at this time, in which dates are OK).
Why this change between 2021 and now ? Impact of "GPS week number rollover" was in 2019 april, so it was already the case for long in 2021, and nevertheless, there were no impact on my phone at this time.

Two facts could explain this change : update of system software on my phone, or update of OruxMaps itself.
First option can be definitely given up : my phone (that is quite old) does not receive any update for long time, last update was in 2018 or 2019...
On the other hand, in 2021, the "free" version of OruxMaps was 7.4.23 or 7.4.24, and we are now in 7.4.26

I don't know what are exactly changes between these versions (they are not documented, change log says only "Bug correction")

So my question is for Orux/Jose, who is the only one who can know : Which "fixes" in 7.4.25 or 7.4.26 (or maybe in 7.4.24) could explain this unwilled evolution ? Could you publish a 7.4.27 that would "repair" it ?

Thanks in advance

PS : I tried to reinstall 7.4.23 over current 7.4.26 (I kept the old APK), but the system rejected the update... maybe because it is a downgrade ?

It's a phone problem.

Apps can't change GPS or how Android reads GPS.

You possibly have a Samsung, right?

It is a problem that has suddenly appeared in many Samsung models, I don't know the reason, but surely OruxMaps has not been the culprit ;)

Read this:

https://www.androidpolice.com/samsung-s7-edge-s8-update-gps/
https://www.xda-developers.com/samsung-gps-fixes-legacy-devices/

Samsung isn't saying anything, but I think they've had a major problem in their software, not anticipating something that should have been fixed long ago.

In the GP version, an option is implemented that replaces the time of the GPS locations with that of the phone. But it is not the best solution. Normally the manufacturer will fix it.

Version 7.4.xx is older than that patch. It's not easy to partially implement that patch, but if the vendors don't fix the problem (which they created themselves), I'll try to implement a patch.


orux

LaurentG

Hi Orux,

thank you for answering.
Yes, I have a Samsung, a Galaxy  A3-2016 to be specific.

BTW, this galaxy model is not mentioned in any of the two documents you send a link to. And I cannot imagine that Samsug may deliver any evolution to this "old" device.
So I'm very surprised that phone's behaviour may have changed between end 2021 and today, without having loaded any system update...

If you say there's nothing in evolutions between 7.4.23 and latest 7.4.26 that could explain this behaviour change, I believe you.
But I definitely do not understand in this case what could have generated it.

And since I'm (almost...) certain that Samsung will never provide a fix for A3-2016, I just hope that you'll eventually implement the patch you mention...  :-*

And if not, it's not absolutely "dramatic" : Error on date/time is "only" in Segments (columns segfechaini and segfechafin) and TrackPoints (col. trkpttime) tables, but date/time is correct in Tracks (col. trackfechaini) and in Pois (col. poitime) tables (likely because in these tables it is taken from phone time, and not from GPS time ?)

Have fun !

orux

Quote from: LaurentG on September 12, 2022, 09:33:34 PM
Hi Orux,

thank you for answering.
Yes, I have a Samsung, a Galaxy  A3-2016 to be specific.

BTW, this galaxy model is not mentioned in any of the two documents you send a link to. And I cannot imagine that Samsug may deliver any evolution to this "old" device.
So I'm very surprised that phone's behaviour may have changed between end 2021 and today, without having loaded any system update...

If you say there's nothing in evolutions between 7.4.23 and latest 7.4.26 that could explain this behaviour change, I believe you.
But I definitely do not understand in this case what could have generated it.

And since I'm (almost...) certain that Samsung will never provide a fix for A3-2016, I just hope that you'll eventually implement the patch you mention...  :-*

And if not, it's not absolutely "dramatic" : Error on date/time is "only" in Segments (columns segfechaini and segfechafin) and TrackPoints (col. trkpttime) tables, but date/time is correct in Tracks (col. trackfechaini) and in Pois (col. poitime) tables (likely because in these tables it is taken from phone time, and not from GPS time ?)

Have fun !
I think Samsung has had a major problem with some library that it uses to manage GPS data on all devices for years now. That's why the patch that affects so many old and not so old models.

From what I have seen in recent years, the fault begins at not very specific moments in time, not exactly in 2019. Since that year they have been informing me promptly that other phones began to have that fault, but from a moment determined in the year 2020, or 2021 or the present year. Always other manufacturers/models. Samsung started a short time ago.
OruxMaps only uses the clock time when you start recording the Track, to give the default name to the Track. Or when you create a waypoint, it uses the clock time at that moment, not the GPS because there may not be recent GPS data.


orux