To jump on the bandwagon, I have made a Gitlab CI builder. This uses the “Devel” region on the code, which means one can combine countries, for example create data for the DACH (Germany, Austria, Switzerland) region.
Just add your locations to the LOCATIONS variable and run the job.
There is also a CI template you can use in your own projects:
Since I have no SDK etc. installed, I had to check how to install RPMBUILD on SailfishOS and had to prepare a SPEC file… only some line of code. But took me some time 
Thought about that, too.
But did/do not think that openrepos is the platform to distribute such huge amount of binary data!?
I made some enhancements to the code to compile whole continents (incl. islands, extreme points). As I do not see the benefit in having only a particular country in my DB. Especially in Europe you travel (okay, not right now 
) so fast that you cross borders quite fast and need all countries combined. Makes more sense imho. And calculation time is just a few seconds more…
But Europe DB is 137MB and whole world DB 312MB.
And that would need to be updated every few weeks/months…
Looking at whole world DB - that is not that much. Especially if moved to /home partition.
Distribution could become an issue. It is a manual process to update packages at OpenRepos and maybe it is better not to stress the bandwidth for it either with larger packages. I wonder if some public p2p swamp can be created to distribute the latest version. Ideally, something that can be joined for an update on a phone and then dropping out after that. Having that and on top some user’s PCs keeping master copies, should make distribution possible and without any major strain on OpenRepos.
I think that it would be better to have an app (settings page?), which update/downloads and pipe the bit stream directly into the converter. Could be easily done even with a Bash script, if we would have Bash by default.
Hello cartron,
The article looks great! However, I don’t think using Mozilla or Yandex will help to get a quicker GPS fix. I might be wrong, but you need an assisted GPS service for this which I think there is no such implementation in SFOS. On Android, Google is providing all the information from an A-GPS server and GPS fix is acquired quicker.
Here are some details about A-GPS: https://en.wikipedia.org/wiki/Assisted_GPS
That said, I think that the fix on Sailfish OS is acquired only by “searching the sky” and getting the almanac data directly from the GPS satellites.
This takes almost 12.5 minutes the complete as almanac is transmitted by each satellite in 25 frames total, as described here:https://en.wikipedia.org/wiki/GPS_signals#Almanac
If I understand correctly it gets a bit faster if you use the GPS regularly, because the almanac is valid for a couple of days; this must be why some claim it’s not a problem for them.
I wanted to rpmbuild myself, having downloaded csv from OpenCellID, just to test the difference, but having an issue - the latest version is 0.1.5, but build fails, because the script looks for 0.1.2 folder and zip in soirces. So, I tried to use v. 0.1.2, but the script looks for file called “geoclue-provider-mlsdb …”, but the download is called “geoclue-providers-mlsdb …”, (note the ‘s’) . I’ve tried modifying files and folders, but not effective, since the files are created in /var/tmp upon executing the rpmbuild command, (a folder gets written to “BUILD” directory called “geoclue-providers-mlsdb…”). Any ideas how to get around this?
can’t really answer. I’ve built github source in desktop ubuntu 20.04 vm with no problems.
thanks, I’ll update the article
that’s half true - using MLS / cell towers info, you get a “rough” fix (~500 meters for me), and then of course to get an accurate fix, you need GPS and to be outside ideally
So there is really no collaboration between MLS and the actual GPS positioning?
I guess that explains why i got a “fix” 1km away, which then changed to 2km away, and only after said 12.5 minutes outside i got a precise position. This was also lost within a minute of coming back home, and then i was back to 2km away.
An “out of the loop” question:
While community providing solutions fast is AWESOME isn’t there a way for jolla to solve this instead of having to rely on Openrepos packages??
okay, thanks … Could you share your rpm?
Does the MLS Offline Finland contain only the Finland-data or is the separate Åland Islands -data also included?
For those who want to know why it would be relevant, Åland Islands is a demilitarized autonomous region of Finland and it would be nice to have both datasets in one package in case you visit there.
I think Jolla had a solution but had to stop using it because of a patent troll, which is why they could no longer provide the service.
ok, I got my app to build with qmake/make. Working now, but i’m not sure of the proper syntax with multiple mcc #s, in the source code edit… If I do ‘&& mcc==310-316’, it finds 0 cell towers. If I do ‘&& mcc==310’, it finds 5, ‘&& mcc==311’ finds 600,000… What’s the code to add all from 310-316, or does it merge data if i run it with different mcc? (update - I tested and it doesn’t merge, but overwrites)
Thanks for your help…
What about
‘&& (mcc>=310 && mcc<=316)’