[mkgmap-dev] [PATCH v1] Multipolygon: do not warn about joining problems for ways outside the bbox

Wed Nov 10 07:28:10 GMT 2010

Hi aighes,

On Tue, Nov 09, 2010 at 05:27:28PM -0800, aighes wrote:
>I generated my map with the loging. It only tells me, what
>Päijänne-multipolygon has an error. In josm I can't find an error. Validator
>too, so I think it's correct. In the screenshot you can see, that the
>problem might caused by tile-border.

What are the tile borders? You did not include the areas.list, or any 
splitter output other than the tile file (which Osmosis cannot process, 
as it is in the obsolete 0.5 format). In that file, I see this:

<bounds minlat='59.8974609375' minlon='25.3125' maxlat='63.0615234375' 
maxlon='27.24609375'/>

For what it is worth, until October 24 I had split Päijänne exactly at 
lon=25.3125. I moved the tile border west to 25.157318 in order to get 
all of Päijänne in the same tile (63240003 on my map).

Your mkgmap.log.0 cannot be complete, can it? I only see this line 
there:

2010/11/10 02:01:36 WARNUNG (MultiPolygonRelation): 12000022.osm.gz: 
Multipolygon http://www.openstreetmap.org/browse/relation/1239779 
contains errors.

This should be followed by information on the faulty way IDs, something 
like this (this is a known broken multipolygon that is not rendered by 
the default style):

2010/11/09 12:27:33 WARNING (MultiPolygonRelation): finland.osm.pbf: 
Cannot join the following ways to closed polygons. Multipolygon 
http://www.openstreetmap.org/browse/relation/384611
2010/11/09 12:27:33 WARNING (MultiPolygonRelation): finland.osm.pbf: 
- way: http://www.openstreetmap.org/browse/way/84534115
2010/11/09 12:27:33 WARNING (MultiPolygonRelation): finland.osm.pbf: 
- way: http://www.openstreetmap.org/browse/way/84534118

mkgmap should warn about numerous other things, such as unsupported 
day_on, day_off, hour_on, hour_off in turn restriction relations.

Please include your full mkgmap.log.0 file, or at least include the way 
ids that mkgmap is reporting for the multipolygon.  Apparently you have 
filtered the log through grep -vf logging.ignore, using my 
logging.ignore patterns.

I guess that if you are reluctant to move the tile border and if this is 
not fixable in mkgmap, this issue could be fixed by increasing the 
splitter 'radius', so that more 'safety margin' of ways and nodes would 
be included in the tile file.

The purpose of the logging.ignore is to display a summary of errors and 
warnings that will have to be dealt with, on the standard output.  
Because there are known broken multipolygons in the input data, I do 
weed out the detailed diagnostics of all broken multipolygons. If 
logging.ignore lets a multipolygon warning through, I will look at the 
mkgmap.log.0 file for the details.

	Marko