October 2012 - mkgmap-dev - The mkgmap lists

Fwd: Maps randomly disappear when zooming (was Fwd: [OSMPH GPS:25] testing latest mkgmap compiler r2340)
by Steve Ratcliffe 29 Oct '12

29 Oct '12

Hi I accidentally deleted this post from the mailing list moderation queue - sorry. I've removed some of the images that made the mail too big, since the first few demonstrate the point very well. Has anyone seen this problem recently? It looks similar to a very old problem that I've not seen for a long time. ..Steve -------- Original Message -------- Subject: Maps randomly disappear when zooming (was Fwd: [OSMPH GPS:25] testing latest mkgmap compiler r2340) Date: Wed, 24 Oct 2012 11:57:13 +0800 From: maning sambale <emmanuel.sambale(a)gmail.com> To: Development list for mkgmap <mkgmap-dev(a)lists.mkgmap.org.uk> Dear everyone, Sharing this bug report from a user. AFAIK, this affects nuvi 24xx versions. See the attached images. ---------- Forwarded message ---------- <snip> On my Nuvi 2450: it's annoying that portions of NCR erratically disappear at different zoom levels (on both development & the regular map). <snip> -- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------

3 2

Question regarding overlap in splitter and mkgmap
by GerdP 29 Oct '12

29 Oct '12

Hi, I am again working on an algorithm for splitter that detects all incomplete ways and rels. It works fine with input like finland.osm.pbf (~100MB) and -Xmx1000M, but it probably would fail to process europe on a 32bit machine. Up to now, we use the overlap value to reduce the likelyhood of incomplete ways and rels. Now, I wonder what exactly splitter should do with a way that has only points in this overlap. To make it easier, let's assume that this way is not a member of a relation. Is such a way used in mkgmap or is completely ignored? If it is completely ignored, I think splitter should not to write it, else it should try to write it with all nodes. In other words: If splitter could make sure that every way and relation that a) has a node within the real bbox b) may enclose the real bbox c) may cross the real bbox is written to the tile with all related elements (nodes of ways, members of relations), would we still need the overlap? I remember an algorithm in mkgmap that searches the nearest city. Would this algorithm "see" a city that is in the overlap ? ciao, Gerd -- View this message in context: http://gis.19327.n5.nabble.com/Question-regarding-overlap-in-splitter-and-m… Sent from the Mkgmap Development mailing list archive at Nabble.com.

3 2

Commit: r2348: style-functions branch has been merged to trunk
by svn commit 28 Oct '12

28 Oct '12

Version 2348 was commited by wanmil on 2012-10-28 21:18:10 +0000 (Sun, 28 Oct 2012) style-functions branch has been merged to trunk

1 0

Commit: r2347: Merge the style-functions branch.
by svn commit 28 Oct '12

28 Oct '12

Version 2347 was commited by wanmil on 2012-10-28 10:08:06 +0000 (Sun, 28 Oct 2012) Merge the style-functions branch. As a first function length() is implemented that returns the length of way in m.

1 0

Character /
by RocketMan 28 Oct '12

28 Oct '12

When the character / is used in street names it shows on the map but not in turn directions. Street "Main 234/5" becomes "Main 234".

1 0

Re: [mkgmap-dev] [mkgmap-svn] Commit: r2341: Bump the style version to 1 and use the new include directive for all the address parts that
by Steve Ratcliffe 27 Oct '12

27 Oct '12

On 26/10/12 23:38, svn commit wrote: > > Version 2341 was commited by steve on 2012-10-26 23:38:41 +0100 (Fri, 26 Oct 2012) > BRANCH: style-include > > Bump the style version to 1 and use the new include directive for all the address parts that > are common to the points, lines and polygons files. I'd like to merge back the style include branch soon. I don't want to step on any toes in the style rules, but I've used the include mechanism for all the street address rules that are common across the three element files. Its a good demo, and I think uncontroversial. If you use the include mechanism, you can set the version to 1 instead of 0 so that it will fail on older versions of mkgmap that do not understand 'include'. Well it will fail anyway :) but the error message may be more useful... ..Steve

2 1

[PATCHES v1] Perfomance and memory improvements
by WanMil 27 Oct '12

27 Oct '12

Hi, attached are 5 patches with slightly performance improvements and noticable max memory reductions. free_mem_earlier_v1: The coords map is released before the hooks are started. The coords are used only while the input data is loaded. Additionally all hooks release their data at the end of their end() method. I measured the maximum memory usage by compiling a big tile and minimizing the -Xmx parameter so that no OutOfMemoryError occurred. Using the patch I could compile the big tile with 9% less memory. free_nodes_ways_earlier_v1: All OSM nodes and ways were removed only after they have been completely converted. This patch releases each OSM object just after it has been converted to the intermediate MapElement object. tags_v1: This patch removes some autoboxing (Integer to int) from heavily used methods. I don't know if there still is a penalty for autoboxing with current JREs. But it feels better not to use autoboxing. polyline_addpoints_v1: Instead of adding point by point to a polyine the patch adds full list of points. Arraylists containing the points need to be resized once only instead of multiple times. clipping_v1: The makeBoundaryNodes method clips all ways to the bounding box. But it has to workout all nodes if the way is completely withing the bbox. The patch avoids that. WanMil

1 0

Commit: r2345: Merge across the missing files just added to the style-include
by svn commit 27 Oct '12

27 Oct '12

Version 2345 was commited by steve on 2012-10-27 19:38:36 +0100 (Sat, 27 Oct 2012) Merge across the missing files just added to the style-include branch.

1 0

Commit: r2343: Merge the style include branch.
by svn commit 27 Oct '12

27 Oct '12

Version 2343 was commited by steve on 2012-10-27 19:30:47 +0100 (Sat, 27 Oct 2012) Merge the style include branch. In a style file you can have a statement to include another file. For example: include 'inc/address'; You can also include a file from another style. This is only really useful to include from a built in style. include 'inc/address' from default; The version number of the style format has been updated to 1.

1 0

From where are these ways being labelled?
by Dave F. 26 Oct '12

26 Oct '12

Hi I'm using ver. 2340. In my lines file for the style I'm using I have these options: highway=footway & foot=yes [0x16 resolution 22] highway=footway Using this command line I make a gmapsupp file & copy to my Garmin Oregon 300 "C:\Program Files\Java\jre7\bin\java" -Xmx512M -ea -jar mkgmap.jar --remove-short-arcs --description="Bath & Surrounds" --gmapsupp "Contours.img" --generate-sea:polygons --transparent --tdbfile --style-file="C:\dwgs\Programs\GPS All\mkgmap\Garmin Profile" --family-id=42 M000002.TYP bath.osm In both the GPSr unit & in Mapsource any ways tagged as above are labelled as Public Footpath & Footpath respectively. I can see no reason for this. I've searched all the files in my style profile & find no occurrence of "Public Footpath". Where is it picking up the configuration to amend the labelling? Also, as an aside, why does adding --version option prevent the command from compiling? Cheers Dave F.

5 4