Messages

i do not have cleared the browsercache. when you download files again - it is no question about the cache. you sent new order and get request from the server. cache is a reason, when you have problems with the site and its scripts. here you load the files from servers.

but - i will try it again and let you know what happend. i start this tomorrow because i have to drive my truck now

Hi,

Usually you are right, on openandromaps all browser caching ist disabled in the W3TC plugin to avoid problems with browsers.
However, there are some browsers with exotic settings enabled that may cause confusion.
An other issue is that downloads are numbered if you have one with the same name in your download folder.
So if you have the germany.zip downloaded the "update"-download will have the name germany(1).zip

My recommendation is to clear both download history _and_ the physical downloadfolder bevore updating a map.

How I could get the error message ("polygone is not a polygone")?

Hi,

Redirect the screen output of the mapsforge writer to a file, according to your favorite OS.
For Windows (DosCMD) this is eg. "mapit.bat 2>log.txt" (important is the "2" in the redirect)
Here you will find all the output of the Mapsforge writer.

Eg. Scandinavia produces a lot of errors.
However, OSM-Inspetor is the tool of my choice.

Best regards, Christian

Maybe I did not understand what you have changed. The mapsforge plugin only or together with the writer (for Locus)?
At last it would be nice to make Locus more failure tolerant too.

Hi jusc,

Locus is perfect OK, the problem is related to the mapsforge library (most likely to the writer) this affects all APPs based on Mapsforge.
The logfile of the writer is full of "polygone is not a polygone (or something like this)".
It would be perfect if someone solves the self backing up lines in the Mapsforge Mapwriter in the current Version 3.0 (I will stay with this version as long as possible), all other problems can be seen at the OSM-Inspector.

There has been fix for multipolygon in OSM by christiank61: http://www.openstreetmap.org/browse/changeset/17142409 (Than you!)
I made new vector map for myself today and Locus fill the forest correctly (like Mapnik does).
Now I know where to hunt for bugs... But it seems that OSM is full of bugs like this, where Mapnik is happy and mapsforge is unhappy and skip the object rendering (if I understand correctly).

Glad to help you! (christiank61 is my nick at OSM)  8-)
The only problem not covered by the OSM-Inspektor are the self-backing-up lines, these have to be seached by hand.

Here is a link: http://www.openstreetmap.org/?lat=50.72 ... 6&layers=M
I know that I touched this multipolygon (around "Plane pod Jestedem) in OSM (by using Potlach2) when it stopped working.

Hi Milan,

I corrected the problem (at least the intersection), to be seen at the OSM-Inspector.
As I wrote in my previous post the Renderer of the OSM-Onlinemaps is  most forgiving, even for worst mapping errors - Mapsforge (the Library behind 95% of all Vevtormaps) is most strict.
One hint: try to move from Potlach to Josm, Potlach is a fine Editor and easy to use - however it is responsible for most of the double nodes and selfbacking up ways cause it trys to help the mapper too much in some ways.

yes I see, I tested it with newest Czech map from Openandromaps, because I feared, that the Locus store maps are "outdated". But I get the same screen. I write an email to christian (kech61).

Hi,

Mapnik is forgiving, much too much, so mapper tag their multipolygons, take a look at Mapnik and find that everything is OK, what else should they do...
I'm fighting this problem all the time and just repaired the Thunsee in Switzerland, some coastlines in Scotland, Sardegna --- however, its a fight against windmills  

Take a look at the OSM-Inspector:
http://tools.geofabrik.de/osmi/?view=mu ... tion_lines
The Multipolygon is broken, seems to be an intersection.

If you zoom out you wont even see the map behind all the error-flags of broken MP's

And, hmmmm, even if the MPs are perfect correct (the one in Question is definitly broken) Mapsforge suffers from one more problem:
Some polygones contain an outer member within a inner member (IMO perfect useless but perfect correct    ), Mapsforge cannot handle these situation and skips it.

However, for 99% of skipped MP's the reason ist uncorrect tagging/geometry.

So if you see such a problem, please start the osm inspector + josm and correct the problem, most problems are simple double nodes, self backing up lines (node + forward + node + step back + node_on_node + forward) or intersections.

Hopefully the next version of Mapsforge will contain some code to correct at least the self backing up lines, someone at the mapsforge_dev_group had contributed such a code_snippet.

ah, damn ... you won, for now
and thanks to this, we should change intervals in vector maps and optimize more critical levels 11, 12

If you mean to modify the default base-zoom-levels and add one (or some) - I use this in the Netherlands, France and GB map.
It speeds up the rendering if the tagmapping is carefully adjusted to these base levels.
However, the problem with the original Mapsforge implementation is that the detail reduction applies for all zoom-level that resides within a base range that touches level <12.

So if you have a zoom-interval-conf=5,0,6,8,7,9,11,10,12,15,13,21 level 12 is redered wit simplification (usualy its not) cause the intervall 11,10,12 is lower than the simplification limit of 12

Hmmm.. confusing, I hope you get the picture.

heh, you're around 10 minutes late

viewtopic.php?f=25&t=3189&p=21696#p21696

No, No, NOooo,
I uploaded them 2 days ago !!!
 :twisted:  :twisted:  :twisted:  :twisted:

So the golden banana is MINE

 :mrgreen:

As an alternative suggestion this leads me to a feature request I did mont ago: Different Map sources (vector or bitmap) for each zoom level  :mrgreen: . For example "vector map" on higher zoom and the lower end "bitmap renderings", which looks sometimes better than vector..

Done  

I uploaded a World-Map for both Locus+ all App's capable to handle sqlitedb and one for Orux.
Covered zoom-level: 4-9, bitmap, OSM-Mapnik

[Zoom Level 8 and 9 should always have wood area drawed. It's easier to orient on the card with it.]

Suggested improvment in Openandro:

• Zoom Level 8 and 9 should always have wood area drawed. It's easier to orient on the card with it.

Yes its true - the appearence ist better with wood rendered on level 8 and 9.
However, I hate to see maps slow loading and/or crashing on low powered phones, level 7-9 is most critical in urban areas so I try to avoid huge polygones and all tags not _really_ necessary.
In more remote areas like whole SouthAmerica, Africa, Asia, Russia and Iceland the mapping is much more detailed in lower zoom levels (wood appears on level - thats why I avoid the "zoom-min" in my themes and determine the map-appearence by the tag-mapping while rendering.
There are 7 different tag-mappings for the openandromaps adjusted to the needs of the area maps are covering.

• Amound of village / city naming differs. The locus map naming is better in Zoom Level 8 and 9. Isn't it? (Maybe openandro is better for other resolutions?)[/b]
  

This is an anoying problem with mapsforge, you can do what you want, sometimes the Labels (names) are rendered by chance - the mapsforge group is aware of this problem and will (hopefully) solve it in one of the next versions.

• Hillshade suggestion for Openandro and Locus:[/b]
  There is no hillshade map for locus to download not even a paied one.
  

Hillshading is possible with offline overlays, both Menion and Me have tried this and for me its not the way to go, the amount of data is crazy.
An overlay covering the Alps is as heavy as 9GB (level 5-15), with 8 parallel treads compressing every single tile of a couple of millions on a fast machine it takes days to finally have a size of (still crazy) 3GB.
This tread my find your attention:
viewtopic.php?f=10&t=2923&hilit=hillshade

So I suggest to wait a while till Menion releases his on device hillshade renderer.

You map of 2012 is still better quality with the data included in the zoom levels in comparison to the maps from http://www.openandromaps.org

What do you miss in the maps of openandromaps, I always try to improve them - all ideas and critics are welcome.

[dev_enable_vector_maps_hillshade]

If not, check Locus configuration file http://docs.locusmap.eu/doku.php/manual ... ation_file and last parameter dev_enable_vector_maps_hillshade set to 1. Locus will then use HGT files and generate "shading" on vector maps ...

Hi Menion,

Always a step ahead    , isnt it  
I tried your solution with viewfinder .hgt's and for me this is deffinitly the way to go - thanks for bringing this Beta-feature to my attention, it will save a lot of time for walking, cycling etc. instead of sitting at the computer - THUMBS UP!

Hello Christian,
  I played a while with Maperitive.

Hi Menion,

After some testing with maperative I can say that it works.
You have to set map-background-opacity  to "0" in the ruleset and reaply it right bevor generating Hillshade (the ordinary hillshade, not the Igor-relief)

The problem ist that Mobac destroys the transparency of the tiles when trying to shrink them.
So the Tiles have to generated and processed within the tiles/-Folder one by one, preserving transparency,  _bevore_ feeding them into mobac.

HUHU..  

This means for the Alps processing of about one millon tiles (Level 8 to 15).
I wrote a script for parallelization of this, its curently running, an running and...

The tiles/-folder for the Alps (Level 8-15) is as heavy as 9GB,  :shock: I hope to shrink it to >2GB

For a small area I tested this and it works = maperative>shrinking Tiles>Mobac>RmapsSqlite>Locus, however its real heavy number crunching.

Edit: I tested mbtiles-Format right out of Maperitive and it worked, however there is noch chance to shrink it so: not the way to go.

Hi Menion,

I will stop this (hillshading with Mobac) and switch to Maperitive - I'v read that Locus support MBTiles sqlite-format > is this Format capable to support transparent hillshading as overlay in Locus?
Actually I cant provide an example - I'v to learn to drive Maperitive  :roll:

Best regards,
Christian

Christian, I thanks you ...

anyway I'm quite confused now. File ShadeTest_png8bit.sqlitedb seems to contain 1364 tiles where all are just 141 bytes big images. In all programs I tried, it display just completely black tile. Are you sure this works? Because 141 bytes is quite low on tile where should be at least something ... (and all, tested 3 from different places, are same 141 b)

Hi Menion,

I just tried to work with a original tile (with lots of elevation data) from tileserver and optimized it in Fireworks.
original file: PNG 43k
jpeg50%: 7k
optimized png (palette optimized, Alphatrans, png8, bells n' whistles) = 10-14k

So the png8-file from Mobac is definitly broken and the way to go is IMO the optimized (shrinked) jpg.
I hope that the 50%jpeg is OK out of Mobac.

The overall size of the resulting sqlite-db seems to vary between the different sqlite-formats too.

I will do some further testing and report.