Re: [APP] - version 3.39.+ ( 25. 7. 2019 )

[poutnikl]

@Žajdlík Josef
Perhaps, it may be related to the recent fixed issue of escaping the route planner if Locus was brought back by clicking on its icon, versus staying in route planner if Locus was restored from the list of recent applications.

Sent from my Xiaomi MI A2 / Android 9, via Tapatalk

[balloni55]

V 3.40.0 RC1
where can i find Locus GPS-Averaging ??

[menion]

@tapio
an interesting problem that already reported John few days ago. I may look on it next week, anyway usual problem ... how to simulate it :/

About the content of trackpoint detail: little patience, please. This screen requires complete redesign = planned quite soon I believe.

@Žajdlík Josef
eh ... I made no change regards this problem. Maybe as Libor wrote, it is related to some other fix, but really should not be. The logic here is quite simple.

@balloni55
a) I'm still not satisfied with computed accuracy. The current number is more a fake than real value.

@poutnikl, any quick tip of an algorithm to compute the precision of average value computed from values with own known precision (coordinates & it's HDOP) and that most probably does not have a normal distribution (I consider it has for now to simplify it). The value does not need to be super precise I believe also because HDOP of locations is also an estimate of computed least square method used in core algorithm.

b) most probably, the feature will be included in new Locus Map 4 later. So till then, it will be in regular beta versions only (this one is an exception because it should be really as close to final release version as possible). Sorry for the confusion.

[poutnikl]

Just quick thoughts:

The only idea coming to my mind is error propagation via sums of variances ( squares of standard deviations ).

But to get a model how these propagate via Kalman filtering may be very challenging.

Some info could be extracted from Kalman filtre variance matrices.

Another aspect, that may nullify the above effort would be  nature of position errors.

They are not random either in time, either in space, but there is strong both space and time correlation of errors.

By other words, following a road, there may be e.g. for 10-15 seconds a systematic position bias few meters to the right, and after a short while, a systematic bias to the left. If these biases are long enough, Kalman filtering cannot say difference between the bias and true deviation, as it is designed against random variations.

A filter addressing biases would need a feedback from a real life, like systems evaluating position errors of known positions, what is out of ordinary GPS abilities.

Sent from my Xiaomi MI A2 / Android 9, via Tapatalk

[Viajero Perdido]

Trying the RC...  I'm okay with the lack of built-in GPS averaging, and it's nice to see the app still uses the external version if available.  (I kept a copy of 3.0 which is pre-malware.)

But best of all, after much couch-surfing, I see the issue of cut-off texts with MFV4 maps is almost completely gone, as I probably mentioned before.  It still happens, but very rarely, maybe 1% as often as it used to.

Map rendering does seem a bit slower, BUT as I think I'd said, I'm perfectly willing to accept that in exchange for proper map rendering; I'm guessing it needs to re-render some tiles to fix the text.  Also, I'm testing the RC on an older, slower device, so it's not really a fair comparison.  In short, speed is okay, rendering is wonderful.

LoMaps still seem to be V3.  Will they switch to V4 at some point?  (Sorry if it was mentioned somewhere; I haven't seen it.)

[john_percy]

@menion @tapio
This is the reference I found to this error or similar in Mapsforge: https://github.com/mapsforge/mapsforge/issues/929#issuecomment-284289541
(See issue #1)

However Locus very rarely chooses to render labels as multi line text.

[Andrew Heard]

@Andrew Heard
I've set export to directory Locus/export/test, then I've deleted "test" directory in file browser and then pressed "export" button. And no error, all went well. "test" directory was automatically created and inside is exported file. So please, precise steps or video needed if you want to fix it, thanks.

@menion - 3.40.0 RC1 - I recorded a video https://www.youtube.com/watch?v=OmI-tEw-t5E of the very brief error msg related to a track export to an invalid directory - see at the 2 second mark.

[menion]

@poutnikl
thanks Libor, so as I expected, no simple solution. I completely forget on Kalman filter that may be enabled in the time of "averaging" ...

@Viajero Perdido
I'm always glad to read some positive news about recent changes, perfect.
Performance is a little worst, it is a correct observation. Anyway, I see there still few places where performance may be improved so it should be better over time.

Switch of LoMaps to V4 is in road-plan, but not with highest priority.

@Andrew Heard
ah I see it now, thanks for the video. It is an issue caused by incorrectly defined directory in app main settings > Misc > Default directories > set "export" directory. You maybe wrote this before. I still thought that it is incorrectly set an export directory for this single track.

I made tiny improvements here just to display better notification info. Thanks

[lor74cas]

Today I recorded a track for less than 2h with last beta, Locus used 10% of battery power, it's the first time I see it so hungry.
In this time Locus worked in background and the phone was untouched in my pocket.

Inviato dal mio SM-A520F utilizzando Tapatalk

[menion]

5% per hour? This is amazing, I want this too
(I've usually around 10%)

[lor74cas]

5% per hour? This is amazing, I want this too
(I've usually around 10%)

Ask the developer to reduce it on your phone

Inviato dal mio SM-A520F utilizzando Tapatalk

[menion]

Battery consumption is affected by too many parameters to simply say reason for your increase.
It is needed to have constant, repeatable conditions.

I've just published one more Beta version 3.40.0 RC2. Among the few minor fixes, I've spend almost two days on speed up of the cold start itself. Little risky just before release so hope, there won't be any serious issue. (Michal reported me a problem with loading some LoMaps, but I have no more information from him, so hope it was unique problem).

Enjoy Sunday.

[erfi]

Beta version 3.40.0 RC1 + Beta version 3.40.0 RC2: Database Menu Track: When saving an already existing track, the app crashes. The thumbnail will also be lost.

[menion]

"Saving an already existing track" > give me please exact steps to reproduce. Edit over Route planner does not cause a crash to me. Thanks.

[Viajero Perdido]

I've spend almost two days on speed up of the cold start itself.

Holey moley, that's fast now!  No problems seen with RC2 on my Android 6 / Nexus 7 with its motley collection of maps, points, etc.

It was only a quick test, all I had time for.  Cheers..