[mkgmap-dev] How to track an error in OSM data ? (and style's behavior error)

Sun Jul 31 11:32:01 BST 2011

> Hello,

Hi David,

>
> I compiled r2009 with PBF format support and when I compile (with r2008
> & r2009) Ireland PBF data from geofabrik I have a corrupted map at high
> detailed level (24) in Basecamp.

What do you mean with "corrupted map"? Does Basecamp crash? Do you miss 
some data or what else?

> The problem occurs with at least one
> "commercial" polygon in Belfast area. Some huge polygon are displayed
> above land and some sea parts. Do you know how I could find the bad
> polygon in source data (in OSM format, it is far more easier to read it
> with human eyes) ? Or is it a bug well defined ?

Mmmh, I don't know an easy solution.
Search for one huge polygon which covers some sea parts. The huge 
polygon may have one ore two long straight lines. These may be 
artificial lines created by mkgmap. Look at the point where the straight 
line begins and search in the real OSM data for a polygons or 
multipolygon which is located at this point.

By the way: Do you use the tile splitter to create the tiles?

>
> I also found an error in a style file for forest polygons. You have to
> write the rules like this if you want the result expected :
>
> landuse=forest & wood=coniferous [0x54 resolution 18]
> landuse=forest & wood=decideous [0x50 resolution 18]
> landuse=forest & wood=mixed [0x55 resolution 18]
> landuse=forest & wood !=* [0x50 resolution 18] /*this line is applied if
> no data matches the ones above*/
>
> and not like that :
>
> landuse=forest & wood=coniferous [0x54 resolution 18]
> landuse=forest & wood=decideous [0x50 resolution 18]
> landuse=forest & wood=mixed [0x55 resolution 18]
> landuse=forest [0x50 resolution 18]/*this line overrides the ones above*/

I am not a style guru. As far as I know all rules after the first 
matching line with a garmin id are skipped unless you add a continue 
statement. So the default style should be correct.

>
> David

WanMil

>
> PS : well done with country/administrative data.

Thanks! It's fun!