June 2015 - mkgmap-dev - The mkgmap lists

Commit: r3620: improve throughput in HousenumberGenerator
by svn commit 10 Jun '15

10 Jun '15

Version mkgmap-r3620 was committed by gerd on Wed, 10 Jun 2015 improve throughput in HousenumberGenerator - shuffle road points before adding to (unbalanced) k-d-tree, this drastically reduces tree depth and thus improves both the creation speed and the search speed - create only one k-d-tree The positive effect is rather small in areas like Germany, rather big in Africa (many long roads, few housenumbers)

3 2

Commit: r3619: improve splitting of long road segments to prevent assertion "arc segment with big bbox not split"
by svn commit 10 Jun '15

10 Jun '15

Version mkgmap-r3619 was committed by gerd on Wed, 10 Jun 2015 improve splitting of long road segments to prevent assertion "arc segment with big bbox not split" The old code calculated a fraction value which did not always lead to a segment which was short enough, probably because one routine uses flat earth approximation and another doesn't. For long way segments the flat earth approximation is not appropriate.

1 0

Commit: r3618: housenumbers: fix possible "internal error, house has wrong segment .. " message
by svn commit 10 Jun '15

10 Jun '15

Version mkgmap-r3618 was committed by gerd on Wed, 10 Jun 2015 housenumbers: fix possible "internal error, house has wrong segment .. " message Correct check which detects random order, it did not work for the sequence "4(0), 3(0), 2(0), 1(0), 1(0), 4(0)" case that did not work: the num that is not in order is preceded by two are equal numbers, here the last "4" is preceded by two "1".

1 0

problem in mkgmap
by Steve Sgalowski 10 Jun '15

10 Jun '15

1 africa needs admin level 5 bu unshure which state triggered this then have a prob wher a bbox is not spli only found on compile stephen

2 1

mkgmap
by Steve Sgalowski 09 Jun '15

09 Jun '15

now when i go to split and make a new map for canada , the file list failes meaning , it failes to find all files to complete the map any ideas stephen splitter 429 mkgmap 3617

1 0

Water polygons in default style
by Carlos Dávila 08 Jun '15

08 Jun '15

According to wiki [1] landuse=reservoir and natural=water+water=reservoir are equivalent, but the latter (and preferred) form is missing in default style. Another thing to comment is that all water bodies are treated the same in default style, despite they can vary from a few square meters (ponds) to thousands hectares. I've been playing a bit with area_size() and found some values that avoid too small water polygons at certain zoom levels in MapSource (it may require some more test though). Below is current water_polygons file in my style. Do you think it's worth including such rules in default style? landuse=basin|landuse=reservoir & area_size()>125000 [0x3f resolution 18] landuse=basin|landuse=reservoir & area_size()>35000 [0x3f resolution 20] landuse=basin|landuse=reservoir [0x3f resolution 22] natural=bay [0x3d resolution 18] natural=glacier [0x4d resolution 18] natural=marsh [0x51 resolution 20] natural=mud [0x51 resolution 20] natural=wetland [0x51 resolution 20] natural=water & water=reservoir & area_size()>125000 [0x3f resolution 18] natural=water & water=reservoir & area_size()>35000 [0x3f resolution 20] natural=water & water=reservoir [0x3f resolution 22] natural=water & area_size()>125000 [0x3c resolution 18] natural=water & area_size()>35000 [0x3c resolution 20] natural=water [0x3c resolution 22] natural=waterfall | waterway=waterfall [0x47 resolution 21] natural=sea [0x32 resolution 10] waterway=riverbank [0x46 resolution 20] [1] http://wiki.openstreetmap.org/wiki/Key:water

3 5

Rendering of islands
by Albrecht Dreß 08 Jun '15

08 Jun '15

Hi all, I have a (hopefully not too dumb) questing regarding the display of relations with <tag k='place' v='island'/> <tag k='type' v='multipolygon'/> If I simply run 'mkgmap --route --index --code-page=1252 --gmapsupp --tdbfile' on the josm export, the result looks like the left in the attached screen shot (Giglio island, from QLandkarte GT, looks the same in the Garmin). The relation is apparently converted into a Garmin type 0x53 covering the whole island, and hiding all other features. If I copy the default style and just disable (comment out) the 'island' and 'islet' polygons, the result is the right one. However, I feel this is the wrong approach, as now the island's name is not available any more... But what is the correct one? Thanks in advance, Albrecht.

3 3

Commit: r3617: treat natural=water & water=reservoir like landuse=reservoir
by svn commit 08 Jun '15

08 Jun '15

Version mkgmap-r3617 was committed by gerd on Mon, 08 Jun 2015 treat natural=water & water=reservoir like landuse=reservoir Effect of this change: natural=water & water=reservoir will appear at res 20, not 18 - Carlos Dávila

1 0

superfluous country specific rules in inc/address?
by Gerd Petermann 08 Jun '15

08 Jun '15

Hi all, Please check my post http://gis.19327.n5.nabble.com/Wrong-adresses-for-3-french-cities-tp5847140… I don't know why we have so many country specific rules which seem to do the same as the last "general" rules to set mkgmap:city and mkgmap:region. I'd prefer to reduce the country specific rules so that it is clear which countries do not use mkgmap:admin_level8 for city and mkgmap:admin_level6 for region. Comments? Gerd

4 12

Commit: r3616: improve address search for Belgium
by svn commit 07 Jun '15

07 Jun '15

Version mkgmap-r3616 was committed by gerd on Sun, 07 Jun 2015 improve address search for Belgium - by Minko

1 0