October 2009 - mkgmap-dev - The mkgmap lists

tweezing-arcs: success
by Valentijn Sessink 19 Oct '09

19 Oct '09

Mark, list, Today I had the chance of biking with an "arc tweezed" map on a route that went wrong, previously. The tweezing really makes routing much, much, much better. There is of course the occasional why-do-you-send-me-this-silly-way routing decision, but generally, bicycle routing has suddenly improved to a useful state; some of the routing decisions are probably even the result of lacking "cycleway=opposite" tags. Really good. Occasionally, there was an odd direction, but that, too, was probably the result of mapping oddities. So, thanks again for the wonderful job mkgmap.jar does :) Valentijn

1 0

Commit: r1294: Added duplicate() method.
by svn commit 19 Oct '09

19 Oct '09

Version 1294 was commited by markb on 2009-10-18 21:56:03 +0100 (Sun, 18 Oct 2009) Added duplicate() method. Makes a new way that has the same values (id, name, tags, points) but differs in that the points list is a shallow copy of the original so that subsequent changes to the structure of the original points list don't cause the points list in the duplicate to change (of course, the points themselves could be changed).

4 4

Commit: r1296: Discovered a bit in mdr11 header flags for when there are more
by svn commit 19 Oct '09

19 Oct '09

Version 1296 was commited by steve on 2009-10-19 18:35:42 +0100 (Mon, 19 Oct 2009) Discovered a bit in mdr11 header flags for when there are more than 32k cities. Also prepare for string offsets that need more than 3 bytes, it may well not work...

1 0

is this a known crash
by Apollinaris Schoell 19 Oct '09

19 Oct '09

When --index is used I have 5 of 433 tiles crash with an exception. Can provide the data somewher fro download if someone is interested in debug. Exception in thread "main" java.lang.IndexOutOfBoundsException: Index: 13721, Size: 258 at java.util.ArrayList.RangeCheck(ArrayList.java:547) at java.util.ArrayList.get(ArrayList.java:322) at uk.me.parabola.imgfmt.app.lbl.LBLFileReader.readPoiInfo(LBLFileReader.java:354) at uk.me.parabola.imgfmt.app.lbl.LBLFileReader.<init>(LBLFileReader.java:71) at uk.me.parabola.imgfmt.app.map.MapReader.<init>(MapReader.java:80) at uk.me.parabola.mkgmap.combiners.MdrBuilder.onMapEnd(MdrBuilder.java:108) at uk.me.parabola.mkgmap.main.Main.endOptions(Main.java:358) at uk.me.parabola.mkgmap.CommandArgsReader.readArgs(CommandArgsReader.java:124) at uk.me.parabola.mkgmap.main.Main.main(Main.java:121) -- apo

2 2

Commit: r1295: An mdr file can easily overflow the maximum number of blocks. So:
by svn commit 19 Oct '09

19 Oct '09

Version 1295 was commited by steve on 2009-10-19 15:44:31 +0100 (Mon, 19 Oct 2009) An mdr file can easily overflow the maximum number of blocks. So: 1. increase the default block size to 4096 as there is not too much space penalty for larger blocks in this file. 2. Change the message on block overflow to make it clear that there is something you can do about it by changing block-size option. 3. Make the MDR code actually look at the block-size option:) The new default block size of 4k allows for an mdr file up to 256M At a later time, we may attempt to guess a value for the block size in the mdr code.

1 0

Minor issue with road angles
by Paul 19 Oct '09

19 Oct '09

http://www.openstreetmap.org/?lat=53.71858&lon=-2.62873&zoom=16&layers=B000… I was recently travelling south down the M61 and turned off this to head east along the M65. After joining the slip road from the M61 the directions I were given were to keep right to the roundabout and then take the 1st exit off the roundabout. As you can see there is a feeder lane from the M61 slip road onto the M65 slip road so the roundabout is clearly not the best option. I don't recall this ever happening before so I can only presume it's related to the arc tweezing stuff that was committed recently. I used mkgmap r1280 and the following command line to produce the map java -Xmx2048M -Dlog.config=$abspath/logging.properties -jar mkgmap.jar --adjust-turn-headings --remove-short-arcs --drive-on-left --check-roundabouts --add-pois-to-areas --latin1 --gmapsupp --route --style-file=$stylefiles --max-jobs --code-page=1252 --country-name="Great Britain" --country-abbr=GB --tdbfile --tdb-v4 --description="Map of GB" --family-name="Great Britain" 700000??.osm $contourdir/uk/*.osm >> debug.txt 2>&1 Cheers Paul

3 3

mkgmap splitter --resolution broken?
by Felix Hartmann 19 Oct '09

19 Oct '09

I have problems to understand the implications on memory use of --resolution parameter. It would be logical to think, wouldn't it, that resolution=16 needs less memory than resolution=13 as there are less resolutions to write (24-16 instead of 24-13). However if I run asia.osm.bz2 from geofabrik (and also others) but on asia.osm.bz2 it is noticeable easiest with resolution=13 it needs about 600MB of RAM. With resolution==16 if fails on 4GB, with resolution==17 if fails on 6GB. Is this a bug or intended? Sorry but I don't have time to errortest or provide further information as I'm on the go to London for one week. Just wanted to inform about this behaviour. Felix

1 0

Commit: r1278: Add ability to set a road's class and speed from a CoordPOI.
by svn commit 18 Oct '09

18 Oct '09

Version 1278 was commited by markb on 2009-10-15 09:49:26 +0100 (Thu, 15 Oct 2009) Add ability to set a road's class and speed from a CoordPOI. If the way has a point with a POI that has mkgmap:road-class or mkgmap:road-speed tags they will be transferred to the way along with any mkgmap:road-(class|speed)-(min|max) tags. The tag values will be used to modify the way's class/speed parameters. Furthermore, the way will be split at the points before and after the POI to limit the region that has the class/speed modified. The mkgmap:road-class and mkgmap:road-speed tag values can either be an absolute integer value or they can be prefixed with + or - to indicate that they are relative to the values set by the style rules. The min/max tag values are all absolute values.

4 25

Commit: r1293: Fix regression: don't change roadSpeed if maxspeed has illegal format.
by svn commit 18 Oct '09

18 Oct '09

Version 1293 was commited by markb on 2009-10-18 13:50:30 +0100 (Sun, 18 Oct 2009) Fix regression: don't change roadSpeed if maxspeed has illegal format.

1 0

Commit: r1281: Make an unsigned get3() varient, and use it for quantities
by svn commit 18 Oct '09

18 Oct '09

Version 1281 was commited by steve on 2009-10-15 21:57:56 +0100 (Thu, 15 Oct 2009) BRANCH: mdr Make an unsigned get3() varient, and use it for quantities that are not signed.

3 2