Turn restriction for highway=motorway[_link]
Hello, I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing. Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...? Any hints? Kindly Hendrik Oesterlin ___________________________________________________________ Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...? If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
"Carlos Dávila" cdavilam@orangecorreo.es wrote on 04/08/2010 at 20:14:50 +1100 subject "[mkgmap-dev] Turn restriction for highway=motorway[_link]" :
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...? If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
Thank you for your hint! This is probably the problem. In my styles I have: # motorways highway=motorway {add oneway = yes; add bicycle = no; add foot = no; name '${ref|highway-symbol:hbox} ${name}' | '${ref|highway-symbol:hbox}' | '${name}'} [0x01 road_class=4 road_speed=7 resolution 12] highway=motorway_link {add bicycle = no; add foot = no} [0x01 road_class=4 road_speed=4 resolution 16] I re-used a fileset from the All-In-One server. -- Sincerely Hendrik Oesterlin - email hendrikmail2002@yahoo.de ___________________________________________________________ Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de
On Wed, Aug 04, 2010 at 09:07:27PM +1100, Hendrik Oesterlin wrote:
"Carlos Dávila" cdavilam@orangecorreo.es wrote on 04/08/2010 at 20:14:50 +1100 subject "[mkgmap-dev] Turn restriction for highway=motorway[_link]" :
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...? If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
Thank you for your hint! This is probably the problem.
Wouldn't it be a problem of the source data? In my part of the world, motorway links tend to be entering or exiting the motorway at a very small angle, and are inherently oneway. Can you post a link to a twoway motorway link? Do you suggest that I should remove the oneway=yes from the default style? Best regards, Marko PS: not all turn restriction traffic signs need turn restriction relations. For example, if you exit this parking ramp to the north http://www.openstreetmap.org/browse/way/46098413 you will see a sign that forbids a left turn. I did not add a turn restriction for it, because the way to the left is already oneway (forbidding the turn). Could it be that mkgmap or Garmin is misbehaving when you try to add such a "redundant" turn restriction? PPS: I have been using Google Street View (when available) when in doubt (or when I do not remember why I pressed the Lap button to create a waypoint when recording a GPS trace). It is a welcome addition to aerial imagery and other sources.
El 04/08/10 14:04, Marko Mäkelä escribió:
On Wed, Aug 04, 2010 at 09:07:27PM +1100, Hendrik Oesterlin wrote:
"Carlos Dávila" cdavilam@orangecorreo.es wrote on 04/08/2010 at 20:14:50 +1100 subject "[mkgmap-dev] Turn restriction for highway=motorway[_link]" :
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...?
If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
Thank you for your hint! This is probably the problem.
Wouldn't it be a problem of the source data? In my part of the world, motorway links tend to be entering or exiting the motorway at a very small angle, and are inherently oneway. Can you post a link to a twoway motorway link? Do you suggest that I should remove the oneway=yes from the default style?
In most cases motorway links are separate oneway roads, but, at least in Spain, sometimes both directions share the same road. See way 70904729 in the area below http://www.informationfreeway.org/?lat=38.107106989860675&lon=-6.27248206719...
Best regards,
Marko
PS: not all turn restriction traffic signs need turn restriction relations. For example, if you exit this parking ramp to the north http://www.openstreetmap.org/browse/way/46098413 you will see a sign that forbids a left turn. I did not add a turn restriction for it, because the way to the left is already oneway (forbidding the turn). Could it be that mkgmap or Garmin is misbehaving when you try to add such a "redundant" turn restriction?
PPS: I have been using Google Street View (when available) when in doubt (or when I do not remember why I pressed the Lap button to create a waypoint when recording a GPS trace). It is a welcome addition to aerial imagery and other sources.
On 08/04/2010 04:55 PM, Carlos Dávila wrote:
In most cases motorway links are separate oneway roads, but, at least in Spain, sometimes both directions share the same road.
This is true. But since according to the OSM wiki motorway_link implies oneway=yes these parts should be tagged oneway=no. Mkgmap shouldn't try to fix mapping errors.
On 07/08/2010 08:44, Ralf Kleineisel wrote:
Mkgmap shouldn't try to fix mapping errors.
I totally agree. Far too often the options within mkgmap are being used to attempt to paper over the cracks in the OSM database. Please don't do this - go back to OSM & fix the data. It's so much more beneficial in the long (or even short) term. Cheers Dave F.
El 04/08/10 14:04, Marko Mäkelä escribió:
On Wed, Aug 04, 2010 at 09:07:27PM +1100, Hendrik Oesterlin wrote:
"Carlos Dávila" cdavilam@orangecorreo.es wrote on 04/08/2010 at 20:14:50 +1100 subject "[mkgmap-dev] Turn restriction for highway=motorway[_link]" :
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...?
If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
Thank you for your hint! This is probably the problem.
Wouldn't it be a problem of the source data? In my part of the world, motorway links tend to be entering or exiting the motorway at a very small angle, and are inherently oneway. Can you post a link to a twoway motorway link? Do you suggest that I should remove the oneway=yes from the default style?
In most cases motorway links are separate oneway roads, but, at least in Spain, sometimes both directions share the same road. See way 70904729 in the area below http://www.informationfreeway.org/?lat=38.107106989860675&lon=-6.27248206719...
Best regards,
Marko
PS: not all turn restriction traffic signs need turn restriction relations. For example, if you exit this parking ramp to the north http://www.openstreetmap.org/browse/way/46098413 you will see a sign that forbids a left turn. I did not add a turn restriction for it, because the way to the left is already oneway (forbidding the turn). Could it be that mkgmap or Garmin is misbehaving when you try to add such a "redundant" turn restriction?
PPS: I have been using Google Street View (when available) when in doubt (or when I do not remember why I pressed the Lap button to create a waypoint when recording a GPS trace). It is a welcome addition to aerial imagery and other sources.
On Wed, Aug 04, 2010 at 04:55:42PM +0200, Carlos Dávila wrote:
El 04/08/10 14:04, Marko Mäkelä escribió:
On Wed, Aug 04, 2010 at 09:07:27PM +1100, Hendrik Oesterlin wrote:
"Carlos Dávila" cdavilam@orangecorreo.es wrote on 04/08/2010 at 20:14:50 +1100 subject "[mkgmap-dev] Turn restriction for highway=motorway[_link]" :
El 04/08/10 10:14, Hendrik Oesterlin escribió:
Hello,
I have some trouble to make turn restrictions work properly. I think it is because mkgmap is failing to make them work if the highway types are "motorway" or "motorway_link". The roads are behaving a bit like "oneway=yes" in some directions. A very unpredictable thing.
Is this a known behaviour? I am not jet 100% sure that the issue is really triggered by the "highway=motorway". But maybe someone before has already spend enough time to do all the trial-and-error testing for this...?
If you are using the dafault style, these lines in the lines file may be the key: highway=motorway {add oneway = yes; add bicycle = no; add foot = no } [0x01 road_class=4 road_speed=7 resolution 14] highway=motorway_link {add oneway = yes; add bicycle = no; add foot = no } [0x09 road_class=3 road_speed=2 resolution 16]
In my customized style I have removed add oneway = yes from the motorway_link line.
Thank you for your hint! This is probably the problem.
Wouldn't it be a problem of the source data? In my part of the world, motorway links tend to be entering or exiting the motorway at a very small angle, and are inherently oneway. Can you post a link to a twoway motorway link? Do you suggest that I should remove the oneway=yes from the default style?
In most cases motorway links are separate oneway roads, but, at least in Spain, sometimes both directions share the same road. See way 70904729 in the area below http://www.informationfreeway.org/?lat=38.107106989860675&lon=-6.27248206719...
The default style does "add oneway=yes", not "set oneway=yes". You could fix this by adding "oneway=no" to the two-way section of highway=motorway_link. Actually, I see that you already did that, in the initial version: http://www.openstreetmap.org/browse/way/70904726 Unless you can provide an example where a two-way motorway link that is tagged as oneway=no is misbehaving, I think we should leave the default style as is. (A lazy mapper could forget the oneway=yes, assuming that "everyone knows" that motorway links are oneway, and you could get bogus instructions "leave the motorway at the next entry (not exit)".) Best regards, Marko
El 04/08/10 17:10, Marko Mäkelä escribió:
Unless you can provide an example where a two-way motorway link that is tagged as oneway=no is misbehaving, I think we should leave the default style as is. (A lazy mapper could forget the oneway=yes, assuming that "everyone knows" that motorway links are oneway, and you could get bogus instructions "leave the motorway at the next entry (not exit)".)
I think that is reasonable, so leave the default style as it is now.
El 04/08/10 17:10, Marko Mäkelä escribió:
Unless you can provide an example where a two-way motorway link that is tagged as oneway=no is misbehaving, I think we should leave the default style as is. (A lazy mapper could forget the oneway=yes, assuming that "everyone knows" that motorway links are oneway, and you could get bogus instructions "leave the motorway at the next entry (not exit)".)
I think that is reasonable, so leave the default style as it is now.
Marko, I think you should remove the add oneway=yes. There are a lot of highway=motorway_link that are no oneways and are not tagged with oneway=no. Most mappers do not use oneway=no, because the common understanding of oneway is that oneway=no is superfluous. I know that most of these links should be referenced by two ways, but at the moment they aren't. I have started a quick look around. In each country I needed 1 try only to get a motorway_link without a oneway tag, where it should be oneway=no. France: http://www.openstreetmap.org/browse/way/30069642 Germany: http://www.openstreetmap.org/browse/way/4927295 Luxemburg: http://www.openstreetmap.org/browse/way/23002279 Swiss: http://www.openstreetmap.org/browse/way/57727007 Italy: seems to be ok (I didn't find a quick example) WanMil
On Wed, Aug 04, 2010 at 08:57:42PM +0200, WanMil wrote:
El 04/08/10 17:10, Marko Mäkelä escribió:
Unless you can provide an example where a two-way motorway link that is tagged as oneway=no is misbehaving, I think we should leave the default style as is. (A lazy mapper could forget the oneway=yes, assuming that "everyone knows" that motorway links are oneway, and you could get bogus instructions "leave the motorway at the next entry (not exit)".)
I think that is reasonable, so leave the default style as it is now.
Marko, I think you should remove the add oneway=yes. There are a lot of highway=motorway_link that are no oneways and are not tagged with oneway=no. Most mappers do not use oneway=no, because the common understanding of oneway is that oneway=no is superfluous. I know that most of these links should be referenced by two ways, but at the moment they aren't.
I have started a quick look around. In each country I needed 1 try only to get a motorway_link without a oneway tag, where it should be oneway=no. France: http://www.openstreetmap.org/browse/way/30069642 Germany: http://www.openstreetmap.org/browse/way/4927295 Luxemburg: http://www.openstreetmap.org/browse/way/23002279 Swiss: http://www.openstreetmap.org/browse/way/57727007 Italy: seems to be ok (I didn't find a quick example)
Let's see: osmosis --rx finland.osm --tf accept-ways highway=motorway_link --tf reject-ways oneway=\* --used-node --wx finland-motorway_link.osm yielded 101 ways. I checked some 10 of them, and all should have been oneway=yes. I have no strong opinion on this. If nobody objects, I will remove the add oneway=yes from highway=motorway_link next week. Marko
On Wed, Aug 04, 2010 at 10:19:56PM +0300, Marko Mäkelä wrote:
osmosis --rx finland.osm --tf accept-ways highway=motorway_link --tf reject-ways oneway=\* --used-node --wx finland-motorway_link.osm
yielded 101 ways. I checked some 10 of them, and all should have been oneway=yes.
Found a two-way (at the end of a highway=tertiary) that was not tagged oneway=no: http://www.openstreetmap.org/browse/way/55943238
I have no strong opinion on this. If nobody objects, I will remove the add oneway=yes from highway=motorway_link next week.
Marko
Hi Marko, I was wrong with my feeling that most of the non oneway-tagged motorway_links are oneway=no. Your osmosis call easily shows that. I tested that for some parts of germany. So I suggest to leave everything as it is :-)) Have fun! WanMil
On Wed, Aug 04, 2010 at 10:19:56PM +0300, Marko Mäkelä wrote:
osmosis --rx finland.osm --tf accept-ways highway=motorway_link --tf reject-ways oneway=\* --used-node --wx finland-motorway_link.osm
yielded 101 ways. I checked some 10 of them, and all should have been oneway=yes.
Found a two-way (at the end of a highway=tertiary) that was not tagged oneway=no: http://www.openstreetmap.org/browse/way/55943238
I have no strong opinion on this. If nobody objects, I will remove the add oneway=yes from highway=motorway_link next week.
Marko _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
On 07/08/2010 13:53, WanMil wrote:
Hi Marko,
I was wrong with my feeling that most of the non oneway-tagged motorway_links are oneway=no. Your osmosis call easily shows that. I tested that for some parts of germany.
So I suggest to leave everything as it is :-)) If you know of two-ways that aren't tagged as oneway=no please go to OSM & fix them. It's the best way to get an accurate map.
cheers Dave F.
On 08/04/2010 08:57 PM, WanMil wrote:
I think you should remove the add oneway=yes. There are a lot of highway=motorway_link that are no oneways and are not tagged with oneway=no. Most mappers do not use oneway=no, because the common understanding of oneway is that oneway=no is superfluous.
Motorways and motorway-links in OSM are always implying oneway: http://wiki.openstreetmap.org/wiki/Tag:highway%3Dmotorway http://wiki.openstreetmap.org/wiki/Tag:highway%3Dmotorway_link You do need the oneway=no tag if they are not.
On 04/08/2010 19:57, WanMil wrote:
El 04/08/10 17:10, Marko Mäkelä escribió:
Unless you can provide an example where a two-way motorway link that is tagged as oneway=no is misbehaving, I think we should leave the default style as is. (A lazy mapper could forget the oneway=yes, assuming that "everyone knows" that motorway links are oneway, and you could get bogus instructions "leave the motorway at the next entry (not exit)".)
I think that is reasonable, so leave the default style as it is now. Marko, I think you should remove the add oneway=yes. There are a lot of highway=motorway_link that are no oneways and are not tagged with oneway=no. Most mappers do not use oneway=no, because the common understanding of oneway is that oneway=no is superfluous. I know that most of these links should be referenced by two ways, but at the moment they aren't.
To follow on from what Ralf said; as oneway=yes is implied, removing it will turn them all to two-way. There are far more one-ways than two. The two-way links without the oneway=no are incorrectly tagged & should be amended. Cheers Dave F.
On Wed, Aug 04, 2010 at 09:21:56PM +0100, Dave F. wrote:
To follow on from what Ralf said; as oneway=yes is implied, removing it will turn them all to two-way. There are far more one-ways than two.
The two-way links without the oneway=no are incorrectly tagged & should be amended.
Right. I added oneway=yes or oneway=no for each way in Finland. About five were missing oneway=no and the rest were missing oneway=yes. (Even if oneway=yes is is redundant, it is good for quality control. The search for highway=motorway_link without oneway=* can be repeated, and only any new ways should surface; not the roughly 100 oneway=yes that I added.) I won't change the processing of highway=motorway_link. I just returned from a 175-km bicycle trip on a triplet with two kid stokers (OSM child labor). 30 km of that was on a major highway (trunk?) that is somewhat of a nightmare (no cycleways, heavy traffic, no shoulders) but the only choice in the area. I believe that the default style (correctly) does add bicycle=no to highway=trunk. When I get around to processing my saved tracks, I will add bicycle=yes, foot=yes to the highway, so that Garmin will accept it for non-motorized routing. For most of the trip (where roads existed on the map), I used car routing with shortest route. Best regards, Marko
On 8 August 2010 17:41, Marko Mäkelä <marko.makela@iki.fi> wrote:
shoulders) but the only choice in the area. I believe that the default style (correctly) does add bicycle=no to highway=trunk. When I get
If this is so, then the default should be changed. The fact that various countries have overloaded trunk for their own usage is not a reason to force everybody else to add redundant tags to thousands of km of road. Ultimately I can see this kind of thing beings solved by per-country defaults, but right now such a default just isn't reasonable. Dermot -- -------------------------------------- Iren sind menschlich
On Sun, Aug 08, 2010 at 07:03:33PM +0100, Dermot McNally wrote:
On 8 August 2010 17:41, Marko Mäkelä <marko.makela@iki.fi> wrote:
shoulders) but the only choice in the area. I believe that the default style (correctly) does add bicycle=no to highway=trunk. When I get
If this is so, then the default should be changed. The fact that various countries have overloaded trunk for their own usage is not a reason to force everybody else to add redundant tags to thousands of km of road.
I think that it has to be done on a case-by-case basis. If there is a lesser road nearby, then the default (bicycle=no) is OK. If there is no other practical choice or there is only light traffic on the road, then bicycle=yes makes sense. I would say that bicycle=no is a sensible default for highway=trunk. By the way, at least in Finland, many cycleways are still missing, especially outside cities. There are some cycleways along some trunk roads, close to more densely inhabited areas. Marko
On 9 August 2010 05:56, Marko Mäkelä <marko.makela@iki.fi> wrote:
I think that it has to be done on a case-by-case basis. If there is a lesser road nearby, then the default (bicycle=no) is OK. If there is no other practical choice or there is only light traffic on the road, then bicycle=yes makes sense. I would say that bicycle=no is a sensible default for highway=trunk.
It isn't. bicycle=no means that bicycles are legally precluded from using the road. Nothing on the wiki indicates that this should be assumed and many countries use trunk to tag roads that are both legal and (in many cases) appropriate for bicycles. Dermot -- -------------------------------------- Iren sind menschlich
On 09/08/2010 05:56, Marko Mäkelä wrote:
On Sun, Aug 08, 2010 at 07:03:33PM +0100, Dermot McNally wrote:
On 8 August 2010 17:41, Marko Mäkelä<marko.makela@iki.fi> wrote:
shoulders) but the only choice in the area. I believe that the default style (correctly) does add bicycle=no to highway=trunk. When I get If this is so, then the default should be changed. The fact that various countries have overloaded trunk for their own usage is not a reason to force everybody else to add redundant tags to thousands of km of road. I think that it has to be done on a case-by-case basis. If there is a lesser road nearby, then the default (bicycle=no) is OK. If there is no other practical choice or there is only light traffic on the road
Sorry Marko, but I think you're wrong. bicycle=* should be used to indicate the *allowabilty* (based on law) of bicycles to travel down a way _not_ the *ability* of it. Whether your able to is subjective based on many factors such as surface conditions, usage etc. Cheers Dave F.
On Tue, Aug 10, 2010 at 09:02:11PM +0100, Dave F. wrote:
I think that it has to be done on a case-by-case basis. If there is a lesser road nearby, then the default (bicycle=no) is OK. If there is no other practical choice or there is only light traffic on the road
Sorry Marko, but I think you're wrong. bicycle=* should be used to indicate the *allowabilty* (based on law) of bicycles to travel down a way _not_ the *ability* of it.
OK, I have changed my mind on this, as you could read from my message yesterday. I now think that adding the bicycle=no to highway=trunk{,_link} in r1413 in the default style was wrong. The situation can be somewhat unclear out in the wild too. Technically, traffic signs that would override the default (area-based) maxspeed and access properties should be repeated after every junction, but authorities sometimes omit the signs at minor junctions (bigger than driveways). Especially some sections of cycleways can be officially footways when they forget the bicycle signs. I have been adding bicycle=yes to those that suitable as cycleways by common sense when there is no designated cycleway across the street, and bicycle=designated to those that are indicated by traffic signs. I have also sometimes added bicycle=yes to short sections of footways when it improves the bicycle routing in junctions. Marko
participants (7)
-
Carlos Dávila -
Dave F. -
Dermot McNally -
Hendrik Oesterlin -
Marko Mäkelä -
Ralf Kleineisel -
WanMil