MartinK | # 1361 | 2013-05-05, 21:11 | Report

I've recently fixed some GTK threading issues, so batch download should now be faster and should no longer slow down if there are no GUI changes going on.

Also, I found that the the mbtiles tile storage format is very similar to the modRana sqlite tile storage format. It will not work with modRana as-is, but the changes needed for supporting mbtiles should be quite minor. Basically just telling modRana to look for mbtiles files in layer folder and trying to fetch the tile from them first.

There might be some mbtiles downloads or you can just mill your own tiles with tilemill & then export them in the mbtiles format.

ModRana runs just fine on Linux PCs - just download the tarball (or clone it from github), unpack it somewhere and run modrana.py.

All map tiles will be stored in ~/.modrana/maps

Routing is totally independent from map tiles - the tiles are currently just PNG/JPEG images and don't have any additional use apart from showing the map.

By default modRana uses online routing provided by Google, but also supports offline routing with Monav. See the modRana offline routing guide for how to configure and use it. The modRana Monav routing data repository covers the whole planet, including the recently added Antarctica.

BTW, many thanks to the Masaryk University NLP laboratory for hosting the ~120 GB repository and providing the computational resources that enable global updates taking just 5 hours.

---

Last edited by MartinK; 2013-05-05 at 21:13.

MartinK | # 1362 | 2013-05-05, 21:21 | Report

Some more information about the Monav data repository update

Global update means, that it processes ALL OpenStreetMap data at least SIX TIMES.

• once for big are extracts (Europe, Africa, North America, etc.)
• once for country wide extracts (France, Germany, Austria, australia, etc.)
• 3x as it makes speaprate packs for car, bike and pedestrian routing

And some countries also have sub-regions, so it is even more than 6x times.

And all this (including data download from Geofabrik) in about 5 hours.

The machine has 48 logical CPU cores & ~200 GB RAM. And modRana data repository update can utilize it to 0.90 system load.

---

Last edited by MartinK; 2013-05-05 at 21:23.

malfunctioning | # 1363 | 2013-05-05, 21:33 | Report

I just went through that and it couldn't be any more straightforward. I'm going to go ahead and download the map for my region, then give it a try this week. Thanks!

Regarding Anctartica, how do those directions look?
- Continue 200 Miles and make a left turn on the first penguin.
(205 Miles later)
- Crap, I think we missed that penguin...

malfunctioning | # 1364 | 2013-05-05, 21:36 | Report

Very impressive how the Monav data repository works as well, thank you for sharing that. Open Source FTW!

MartinK | # 1365 | 2013-05-05, 21:45 | Report

Well, I haven't actually tried that yet - I just spotted it got recently added to Geofabrik & thus also processed to the repository.

My guess would be that it just covers the small local road network in some of the bigger polar stations, such as McMurdo.

BTW, this is how McMurdo looks like on OpenStreetMap:

The funny looking roads going over water are IMHO the ice runway access roads.

Estel | # 1366 | 2013-05-05, 21:47 | Report

MartinK, thanks for your incredible work, and congratulations on reaching another platform It's nice to see such wonderful program becoming more widespread, I'm sure, that after becoming fully functional, it just kicks out any Android's "navigational tool" in terms of functionality.
---

Now, out of curiosity - after - at some point in future - bringing full functionality to QML version, are you going to keep working on GTK version, syncing features, etc? It's not a secret, that there are many fans of GTK edition, that doesn't feel QML to be such fancy (not to mention having both GTK and QT bits loaded in memory is a huge bit of resource waste).

/Estel

MartinK | # 1367 | 2013-05-05, 22:04 | Report

Thanks !

I don't have any definitive plans for this but the truth is, that GTK doesn't seem to be doing any inroads to mobile platforms recently and lately everything is Qt.

But due to all the features it has, the GTK GUI is still the main GUI for now and also the main desktop GUI, due to Qt Components missing from most distribution repositories. BTW, a dedicated desktop-like modRana GUI would be nice but I don't see having time for making it any time soon.

Also, I'll be generally improving the core modRana data models, which are shared by both GUIs, so the GTK GUI should also indirectly benefit from QML improvements. So we will see how it goes, at least I want it to remain fully functional and I'll fix any future breakage in the GTK GUI.

Also, I'll welcome any sensible patches for the GTK GUI (and of course for the QML one and generally for any part of modRana)! So if someone wants to improve the GTK GUI while I'm mid-term working mainly on the QML one, now is you chance !

BTW, modRana source code is on Github, so sending a merge request is very easy - just ask Wikiwide or ZogG how easy it is.

---

Last edited by MartinK; 2013-05-05 at 22:23.

magic_doc | # 1368 | 2013-05-06, 06:51 | Report

Hi MartinK,

thanks again for your great work and congratulations for the android-version.

May I dare to ask if there is any chance to get the messed-up gpx-tracks displayed correctly in the future?

CU Michael

Attached Images

 

malfunctioning | # 1369 | 2013-05-08, 00:37 | Report

I've done my first experiments and I'm very impressed! I'm using modRana 0.41.1(GTK version).

A couple of questions:
EDIT: Disregard requests to configure location to save tile data, as I see it has been already asked.

- Is there a way to set the location where tiles are downloaded? I can't find that option anywhere. For that matter, is there also a way to configure where modRana looks for the navigation data?
Please, allow for the configuration of the location to save tiles and navigation data. I would rather put it in the micro SD card because it's easy to replace. I can even put it on a second N900, and use the data as well. Until then, maybe some symlinking is in order.

- When I am setting up the batch download of tiles, I clicked to calculate the size. When it calculates the size, it looks like it keeps calculating the full size (I suppose it downloads and deletes every single tile?) It would be better probably to just estimate by downloading a representative number of random tiles and then extrapolating (sorry if that's the way it already works, I might have not waited enough).

- Another comment regarding the last point: When it is calculating the size, if you click back to exit the menu, the application becomes nonresponsive and needs to be closed. I guess the threads that are downloading / calculating the size of the tiles are still running.

Great program! I think I'm going to use it for the first time tomorrow.

Talking of which, does Google really disable batch downloading of tiles? Because that sucks. Anybody knows how to get around that?

Martin, I just had a cool idea: Since Google disables batch download, but downloading tiles in normal use is allowed, wouldn't there be a way to simulate a route that would cover all the target tiles? Unfortunately I suppose you would have to run the route as many times as the zoom levels you have selected, and change zoom level at each pass. And, the route would probably be huge and hard to compute. A strange benefit is that this would take less space, because it wouldn't download tiles that you wouldn't see during normal use (tiles for places unaccessible within the view distance for the range of zooms selected). Also, you might have to be observant of the speed at which the route is run.

---

Last edited by malfunctioning; 2013-05-08 at 00:55.

malfunctioning | # 1370 | 2013-05-08, 02:39 | Report

Another thing I noticed: There seems to be a discrepancy (at least in some circumstances) between the sizes reported as transferred and remaining on the device, in the batch download page.

I was doing a batch download for Yahoo Maps, and the size available on device was decreasing at a rate 20 or 40 times faster than the size reported as transferred. I also think that the size reported as transferred was too low (around 65MB for 25000 tiles). I'll check the size of the files when the transfer is done, but I thought I'd mention this.