The default style treats motorway_link as oneway by default. I've seen some cases (in the Philippines) where this isn't so. Is it the same case for other countries? If so, I propose we modify the default style. Or, any suggestions to modify the style? http://www.mkgmap.org.uk/svn/wsvn/mkgmap/resources/styles/default/lines -- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------
El 26/10/10 10:53, maning sambale escribió:
The default style treats motorway_link as oneway by default. I've seen some cases (in the Philippines) where this isn't so. Is it the same case for other countries? If so, I propose we modify the default style. Or, any suggestions to modify the style?
http://www.mkgmap.org.uk/svn/wsvn/mkgmap/resources/styles/default/lines
It was discussed some time ago in the list. Conclusion was that most motorway_link are oneway and non oneway motorway_links should be tagged as so in the osm database, so style should remain as it is.
I see, does adding oneway=no to some motorway_link ways override the default? highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16] 2010/10/26 Carlos Dávila <cdavilam@orangecorreo.es>:
El 26/10/10 10:53, maning sambale escribió:
The default style treats motorway_link as oneway by default. I've seen some cases (in the Philippines) where this isn't so. Is it the same case for other countries? If so, I propose we modify the default style. Or, any suggestions to modify the style?
http://www.mkgmap.org.uk/svn/wsvn/mkgmap/resources/styles/default/lines
It was discussed some time ago in the list. Conclusion was that most motorway_link are oneway and non oneway motorway_links should be tagged as so in the osm database, so style should remain as it is. _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------
El 26/10/10 12:06, maning sambale escribió:
I see, does adding oneway=no to some motorway_link ways override the default?
Yes, "add oneway yes" only acts if oneway is not set.
highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
2010/10/26 Carlos Dávila<cdavilam@orangecorreo.es>:
El 26/10/10 10:53, maning sambale escribió:
The default style treats motorway_link as oneway by default. I've seen some cases (in the Philippines) where this isn't so. Is it the same case for other countries? If so, I propose we modify the default style. Or, any suggestions to modify the style?
http://www.mkgmap.org.uk/svn/wsvn/mkgmap/resources/styles/default/lines
It was discussed some time ago in the list. Conclusion was that most motorway_link are oneway and non oneway motorway_links should be tagged as so in the osm database, so style should remain as it is.
maning sambale schrieb am 26.10.2010 10:53:
The default style treats motorway_link as oneway by default. I've seen some cases (in the Philippines) where this isn't so. Is it the same case for other countries? If so, I propose we modify the default style. Or, any suggestions to modify the style?
I checked some examples in Germany: Here it is quite normal, that part of the motorway_links have to lanes, one for each direction. These are either mapped as two separated lines (but mostly they are not separated, so imho this is actually a mapping error), or they are mapped as one single line, but normally without any oneway attribute. What happens in case our default assumption is wrong? - If we assume oneway=yes as default and our assumption is wrong, then some motorway junctions are not usable at all for the routing (either for leaving or for entering the motorway). So the user will be routed to another junction further away. - If we do not assume oneway=yes and our assumption is wrong, the the routing might give some false turn commands (in most cases this will not happen, since the allowed way is also mostly the fastest way), but will always use the nearest junction. As a consequence, I think we shouldn't assume one=yes as default for motorway_links. Gruss Torsten
On 10/26/2010 06:17 PM, Torsten Leistikow wrote:
I checked some examples in Germany: Here it is quite normal, that part of the motorway_links have to lanes, one for each direction. These are either mapped as two separated lines (but mostly they are not separated, so imho this is actually a mapping error), or they are mapped as one single line, but normally without any oneway attribute.
Motorway_link implies oneway=yes according to wiki (http://wiki.openstreetmap.org/wiki/Motorway_link) In case it is not oneway=yes it is a mapping error not to tag it oneway=no. However, I see that this is a very easy mistake to make!
- If we do not assume oneway=yes and our assumption is wrong, the the routing might give some false turn commands (in most cases this will not happen, since the allowed way is also mostly the fastest way), but will always use the nearest junction.
As a consequence, I think we shouldn't assume one=yes as default for motorway_links.
That would work mostly well I guess. However, it is better to fix mapping errors than try to adjust style for correcting them. -- Harri
On Tue, Oct 26, 2010 at 05:17:50PM +0200, Torsten Leistikow wrote:
As a consequence, I think we shouldn't assume one=yes as default for motorway_links.
For what it is worth, when this was discussed last time, most highway=motorway_link in Finland were tagged oneway=yes. Some were tagged oneway=no, and some were missing a oneway attribute. I added oneway=yes or oneway=no to each motorway_link in Finland that was missing it. I invite you to do the same in your country or region. You can easily do that by extracting the ways from a country or state extract with Osmosis and loading the resulting file to JOSM. In JOSM, you can download some surroundings around each line in order to determine whether it should be oneway=yes or oneway=no. Here is a sample command: osmosis --rb extract.osm.pbf \ --tf accept-ways highway=motorway_link \ --tf reject-ways 'oneway=*' \ --tf reject-relations --used-node \ --wx motorway_link.osm josm motorway_link.osm I still do not think that the default style should be changed in this respect, but I don't have a strong opinion, as my "own territory" would be unaffected. I just checked that the above command produced an empty file (just the bounding box). When I replaced the oneway=* with surface=* (I guess many mappers do not bother to set surface=paved on motorways) I did get a number of ways. Marko
On 10/26/2010 09:38 PM, Marko Mäkelä wrote:
file (just the bounding box). When I replaced the oneway=* with surface=* (I guess many mappers do not bother to set surface=paved on motorways) I did get a number of ways.
Motorway and motorway_link imply surface=paved as well. I guess default style could have add surface=paved (if it is not there already). I prefer explicit tags over implied ones though. Perhaps we could mass correct the locally obvious ones like motorway surface. -- Harri
On 10/26/2010 09:45 PM, Harri wrote:
I prefer explicit tags over implied ones though. Perhaps we could mass correct the locally obvious ones like motorway surface.
I think that assuming that motorways are paved and oneway is good. There is no need to add unneccessary tags. Highway=motorway is an abbreviation which makes mapping easier. Otherwise we would not need highway=motorway at all and could map them all as "highway=road; surface=paved; width=8m; oneway=yes; access=motor_vehicles; lanes=2". For motorway_links it is not so clear, but I'd just stick with what's in the wiki because it has been this way for a long time. And errors become quite obvious as soon as someone tries the routing. I haven't seen such an error in a long time around here.
2010/10/26 Marko Mäkelä <marko.makela@iki.fi>:
You can easily do that by extracting the ways from a country or state extract with Osmosis and loading the resulting file to JOSM. In JOSM, you can download some surroundings around each line in order to determine whether it should be oneway=yes or oneway=no.
I did that for all of Germany in early 2009 (i believe), when we had this discussion on this list for the first time(?). It took many hours and the results were: -I got many nice invitations to mapper meetings in various parts of Germany that i've never been to personally -on many motorway_links (that i was aware of) people removed the "unneccessary" oneway=no tags in the following months ;-) -Martin
On Tue, Oct 26, 2010 at 10:23:28PM +0200, Martin Simon wrote:
-on many motorway_links (that i was aware of) people removed the "unneccessary" oneway=no tags in the following months
Did you consider adding note='this is oneway=no because the default for motorway_link is oneway=yes'? :-) Luckily, the motorway_links haven't been subject to an edit war in Finland. There has been some edit war and debate regarding cycleways (which IMO should be tagged as highway=path, segregated=yes/no, bicycle=designated, foot=designated, moped=yes/no, surface=*, lit=yes/no; others want it highway=cycleway, although the ways are almost never marked with a bicycle traffic sign only). As you can see from the above example, I prefer explicit tagging too. It can be a bit overwhelming for beginners, but implied things and country defaults are nasty for anything that tries to interpret the map data. Marko PS: Any objections if the default style adds surface=paved to motorways and motorway links? What does the Vienna convention say about motorways? Does it require a paved surface?
2010/10/26 Marko Mäkelä <marko.makela@iki.fi>:
On Tue, Oct 26, 2010 at 10:23:28PM +0200, Martin Simon wrote:
-on many motorway_links (that i was aware of) people removed the "unneccessary" oneway=no tags in the following months
Did you consider adding note='this is oneway=no because the default for motorway_link is oneway=yes'? :-)
No, and i did nothing to stop the retagging because I didn't want to start and fight in edit wars over single tags in remote areas that i never visited... ;-)
Luckily, the motorway_links haven't been subject to an edit war in Finland. There has been some edit war and debate regarding cycleways (which IMO should be tagged as highway=path, segregated=yes/no, bicycle=designated, foot=designated, moped=yes/no, surface=*, lit=yes/no
Amen. :-)
As you can see from the above example, I prefer explicit tagging too. It can be a bit overwhelming for beginners, but implied things and country defaults are nasty for anything that tries to interpret the map data.
Right, plus, any implied-defaults-list on the wiki can be subject to editwars itself, which causes more damage than edit wars over single explicit tags of single objects in our database.
PS: Any objections if the default style adds surface=paved to motorways and motorway links? What does the Vienna convention say about motorways? Does it require a paved surface?
No, not from my side. IMHO, there is quite a difference between "dictating" global or per-country defaults for highway types and interpreting data when reading it with a tool like mkgmap. In my own mkgmap style, i also try to make "clever" assumptions wheter a way is paved or not and the like, based on the main highway tag, designation, etc. IMHO, the best thing to happen for the database would be tools like mkgmap ignoring top level highway tags to some degree (lets say reducing them internally to pathlike->tracklike->normalroadlike->connectingroadlike->motorwayroadlike) and focusing on the secondary, more specific tags like surface, lanes, width, sac_scale, "grade_seperated" etc. to determine the sub-categories and routing importance. -Martin
On Tue, Oct 26, 2010 at 11:27:27PM +0200, Martin Simon wrote:
PS: Any objections if the default style adds surface=paved to motorways and motorway links? What does the Vienna convention say about motorways? Does it require a paved surface?
No, not from my side. IMHO, there is quite a difference between "dictating" global or per-country defaults for highway types and interpreting data when reading it with a tool like mkgmap.
On a second thought, the default style would not benefit from the surface=paved. There are two rules with { add mkgmap:unpaved=1 } and none that would set mkgmap:unpaved=0, because mkgmap:unpaved=0 is the implicit default.
IMHO, the best thing to happen for the database would be tools like mkgmap ignoring top level highway tags to some degree (lets say reducing them internally to pathlike->tracklike->normalroadlike->connectingroadlike->motorwayroadlike) and focusing on the secondary, more specific tags like surface, lanes, width, sac_scale, "grade_seperated" etc. to determine the sub-categories and routing importance.
That could make sense for motor vehicle routing in densely built areas, yes. Bicycle and pedestrian routing is hopeless beyond distances greater than a few km, because the Garmin algorithm takes unacceptable shortcuts (ignoring all cycleways or paths in the middle of the route). Marko
why cant we change the standard into oneway=no . In big parts of germany only the oneway parts will be tagged mit oneway, but no the toway-parts with onewyy=no. And i think thats the same in other parts of the world ?
On Thu, Oct 28, 2010 at 08:51:49AM +0200, flabot@googlemail.com wrote:
why cant we change the standard into oneway=no . In big parts of germany only the oneway parts will be tagged mit oneway, but no the toway-parts with onewyy=no. And i think thats the same in other parts of the world ?
Better: why can't we change the standard to say that oneway=yes/no has to be specified for highway=motorway_link? A check for that could be easily incorporated in the JOSM Validator plugin. Side note: some months ago, I made Validator not complain about barrier=* ending close to a way. Now that exception seems to have been removed, and I am again seeing those useless warnings. I'm not going to start a editor-edit war on that. :-) Marko
Is it possible to add them in the logging reports within mkgmap? On Thu, Oct 28, 2010 at 3:14 PM, Marko Mäkelä <marko.makela@iki.fi> wrote:
On Thu, Oct 28, 2010 at 08:51:49AM +0200, flabot@googlemail.com wrote:
why cant we change the standard into oneway=no . In big parts of germany only the oneway parts will be tagged mit oneway, but no the toway-parts with onewyy=no. And i think thats the same in other parts of the world ?
Better: why can't we change the standard to say that oneway=yes/no has to be specified for highway=motorway_link? A check for that could be easily incorporated in the JOSM Validator plugin.
Side note: some months ago, I made Validator not complain about barrier=* ending close to a way. Now that exception seems to have been removed, and I am again seeing those useless warnings. I'm not going to start a editor-edit war on that. :-)
Marko _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------
On Thu, Oct 28, 2010 at 03:21:31PM +0800, maning sambale wrote:
Is it possible to add them in the logging reports within mkgmap?
Something like this in the style/*/lines file could do the trick: highway=motorway_link & oneway!=yes & oneway!=no { echo "motorway_link lacks oneway" } Note that normal mkgmap warnings are disabled by default, and you need to specify a log.config parameter to enable them. The "echo" action would probably be always enabled, potentially flooding the log. Therefore I guess I should not add this to the default style, unless there is a strong proposition for it. Marko
Thanks Marko. One more question, the line you added is only for reporting missing oneway tags? highway=motorway_link & oneway!=yes & oneway!=no { echo "motorway_link lacks oneway" } Therefore, I still need the existing style parameters? highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
Note that normal mkgmap warnings are disabled by default, and you need to specify a log.config parameter to enable them. The "echo" action would probably be always enabled, potentially flooding the log.
Yes, I have this enabled.
Therefore I guess I should not add this to the default style, unless there is a strong proposition for it. +1, this should be for debugging only.
-- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------
On Thu, Oct 28, 2010 at 03:45:07PM +0800, maning sambale wrote:
One more question, the line you added is only for reporting missing oneway tags? highway=motorway_link & oneway!=yes & oneway!=no { echo "motorway_link lacks oneway" }
Therefore, I still need the existing style parameters? highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
Yes, the line would have to be specified before the add oneway=yes. Note that I did not test it. You might want to display some diagnostics such as the way id or the lat/lon there, but I do not know how. (Strings such as ${id}, ${lat}, ${lon} would refer to user attributes, not to the internal attributes, as far as I understand.)
Therefore I guess I should not add this to the default style, unless there is a strong proposition for it. +1, this should be for debugging only.
Like I wrote a couple of days ago, for bulk editing, you'd better extract the oneway-less motorway_links from your map extract with Osmosis and loading the result in JOSM. That is, if the above echo rule complains anything, use Osmosis and JOSM to fix it. Marko
Nice! It reported the errors in the terminal: 010/10/28 20:24:59 WARNING (StyledConverter): philippines.osm: Motorway exit Coastal Road Tollbooth (http://www.openstreetmap.org/?mlat=14.48788&mlon=120.98263&zoom=17) has no motorway! (either make the exit share a node with the motorway or specify the motorway ref with a exit:road_ref tag) 38557169: motorway_link lacks oneway 38067665: motorway_link lacks oneway 33043660: motorway_link lacks oneway But I can't see them in the saved mkgmap.log.0.txt . I usually re-visit and share the logs after a couple of days to remove some of the errors. On Thu, Oct 28, 2010 at 4:13 PM, Marko Mäkelä <marko.makela@iki.fi> wrote:
On Thu, Oct 28, 2010 at 03:45:07PM +0800, maning sambale wrote:
One more question, the line you added is only for reporting missing oneway tags? highway=motorway_link & oneway!=yes & oneway!=no { echo "motorway_link lacks oneway" }
Therefore, I still need the existing style parameters? highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
Yes, the line would have to be specified before the add oneway=yes. Note that I did not test it. You might want to display some diagnostics such as the way id or the lat/lon there, but I do not know how. (Strings such as ${id}, ${lat}, ${lon} would refer to user attributes, not to the internal attributes, as far as I understand.)
-- cheers, maning ------------------------------------------------------ "Freedom is still the most radical idea of all" -N.Branden wiki: http://esambale.wikispaces.com/ blog: http://epsg4253.wordpress.com/ ------------------------------------------------------
On Thu, Oct 28, 2010 at 08:42:47PM +0800, maning sambale wrote:
Nice! It reported the errors in the terminal: 010/10/28 20:24:59 WARNING (StyledConverter): philippines.osm: Motorway exit Coastal Road Tollbooth (http://www.openstreetmap.org/?mlat=14.48788&mlon=120.98263&zoom=17) has no motorway! (either make the exit share a node with the motorway or specify the motorway ref with a exit:road_ref tag) 38557169: motorway_link lacks oneway 38067665: motorway_link lacks oneway 33043660: motorway_link lacks oneway
But I can't see them in the saved mkgmap.log.0.txt . I usually re-visit and share the logs after a couple of days to remove some of the errors.
I guess that the echo output goes to the standard output (stdout). In -Dlog.config=logging.properties you could redirect the output to stdout, I guess. In my logging.properties file, there is this definition: java.util.logging.FileHandler.pattern=mkgmap.log If you had all logging on stdout, you could redirect everything to a file, and then grep it. Or you could redirect stdout to another file, say, mkgmap.stdout.txt. Marko
Therefore I guess I should not add this to the default style, unless there is a strong proposition for it.
+1, this should be for debugging only.
-1 as it should be only for debugging purposes. :-) So there is no need to include this in the default style file ;-) But however a good idea for personal style files. Thanks.
Am 28.10.2010 09:14, schrieb Marko Mäkelä:
On Thu, Oct 28, 2010 at 08:51:49AM +0200, flabot@googlemail.com wrote:
why cant we change the standard into oneway=no . In big parts of germany only the oneway parts will be tagged mit oneway, but no the toway-parts with onewyy=no. And i think thats the same in other parts of the world ?
Better: why can't we change the standard to say that oneway=yes/no has to be specified for highway=motorway_link? A check for that could be easily incorporated in the JOSM Validator plugin.
I agree. Motorway Links are very important for routing and any implicit values for oneway are bad IMHO. Chris
participants (10)
-
Carlos Dávila -
Chris66 -
flabot＠googlemail.com -
Harri -
Johann Gail -
maning sambale -
Marko Mäkelä -
Martin Simon -
Ralf Kleineisel -
Torsten Leistikow