access=permissive question

Even cars are routed on this footway: http://www.openstreetmap.org/browse/way/47247295 A default rule in the mkgmap style says highway=footway|highway=path|highway=steps {add access = no; add foot = yes} I think there is a bug in the internal mkgmap code, because the internal code sets this footway to access=yes, so allowing all traffic on this footway. Is this a mappers fault on OSM or a bug in the mkgmap code (committed in r2067)? In my opinion access=yes is a tag that makes no sense at all so it should be deleted anyway, and therefore access=permissive or official should be deleted instead set to yes in the internal codes.

Am 07.08.2012 15:54, schrieb Minko:
Even cars are routed on this footway: http://www.openstreetmap.org/browse/way/47247295
A default rule in the mkgmap style says
highway=footway|highway=path|highway=steps {add access = no; add foot = yes}
I think there is a bug in the internal mkgmap code, because the internal code sets this footway to access=yes, so allowing all traffic on this footway. Is this a mappers fault on OSM or a bug in the mkgmap code (committed in r2067)?
In my opinion access=yes is a tag that makes no sense at all so it should be deleted anyway, and therefore access=permissive or official should be deleted instead set to yes in the internal codes.
I think access=permissive is correctly mapped to access=yes in mkgmap. So the way is open for every mode of transport, cars included. If cars are not permitted foot=permissive is the correct tagging. Chris

Thanks Chris, But does highway=footway not automatically imply foot=permissive? I think mappers tag this and other roads with access=permissive to indicate that this road is open for public and not private. Because access is already set, mkgmap cannot set it to access=no in case of this footway and thus allow all traffic. If all traffic is allowed it is not a footway. So I think mkgmap can better delete or ignore access=yes/permissive tags because they are irrelevant to avoiid inconsequent tagging by mappers.

Am 07.08.2012 18:06, schrieb Minko:
Thanks Chris, But does highway=footway not automatically imply foot=permissive?
hw=footway has default foot=yes. This can of cause be overridden by explicite access tags. access=yes (or permissive) is simply an abbrevation for (foot=yes & bicycle=yes & motorcar=yes & horse=yes, and so on). If you want to be less strict than default mkgmap you can add a style rule. Chris

If I look in the wiki, access = yes only means "the public has an official, legally-enshrined right of access, i.e. it's a right of way" So there is no rule that says all vehicles may use this way. This is set by the highway tag, in this case footway. So only pedestrians may use the way and I don't think this access rule stands above the highway rule. In my opinion access=yes or access=permissive should be removed since it leads to misinterpretation and allows all traffic on those footways. In my styles I will delete them at the beginning of the lines style: access=yes | access=permissive | access=official {delete access}

Am 07.08.2012 19:43, schrieb Minko:
If I look in the wiki, access = yes only means "the public has an official, legally-enshrined right of access, i.e. it's a right of way" So there is no rule that says all vehicles may use this way. This is set by the highway tag, in this case footway. So only pedestrians may use the way and I don't think this access rule stands above the highway rule.
Wiki: Transport mode restrictions Use the access=* key to describe a general access restriction that applies to *all* transport modes. Where different restrictions apply to different modes of transport then mode specific tags can be used. These modal tags each have a place in a hierarchy in which keys become narrower in scope as they branch out from the root.
In my opinion access=yes or access=permissive should be removed since it leads to misinterpretation and allows all traffic on those footways.
In my styles I will delete them at the beginning of the lines style:
access=yes | access=permissive | access=official {delete access}
I think a delete rule for access=yes is already in default style. Chris

On Tue, Aug 07, 2012 at 08:07:30PM +0200, Chris66 wrote:
In my styles I will delete them at the beginning of the lines style:
access=yes | access=permissive | access=official {delete access}
I think a delete rule for access=yes is already in default style.
Yes, since r1985. Any objection to remove other access=* than access=no? I think that access=designated could be removed as well, just in case someone uses it. Marko

Am 07.08.2012 20:57, schrieb Marko Mäkelä:
On Tue, Aug 07, 2012 at 08:07:30PM +0200, Chris66 wrote:
In my styles I will delete them at the beginning of the lines style:
access=yes | access=permissive | access=official {delete access}
I think a delete rule for access=yes is already in default style.
Yes, since r1985. Any objection to remove other access=* than access=no? I think that access=designated could be removed as well, just in case someone uses it.
In my opinion it should be consistent, so either: remove access=yes/official/permissive or be strict and don't remove any of them. Chris

Well, If access=yes was deleted, i didnt had to ask this question. Apperently it was brought back in rev. 2067: Map designatd,permissive,official access to yes. http://www.mkgmap.org.uk/svn/wsvn/mkgmap?repname=mkgmap&path=&peg=2309&rev=2... I agree designated/official/yes/permissive/unknown should be removed Marko wrote:
I think a delete rule for access=yes is already in default style.
Yes, since r1985. Any objection to remove other access=* than access=no? I think that access=designated could be removed as well, just in case someone uses it.
Marko

Marko, The reason that access=permissive/official etc was (incorrectly) hardcoded to yes, lies in this discussion (but it was only about bicycle=*) http://www.mail-archive.com/mkgmap-dev@lists.mkgmap.org.uk/msg10590.html

Chris wrote:
I think access=permissive is correctly mapped to access=yes in mkgmap.
I have contacted the mapper but he doesn't agree with this. access=permissive means that the landowner gave his permission to access this footway. Footways are meant to access by foot only. So no access for all other vehicles.
So the way is open for every mode of transport, cars included.
If cars are not permitted foot=permissive is the correct tagging.
If I map this road to foot=permissive, mkgmap will block this road too: It will add access = no but it can't add foot = yes because there is already a tag foot = permissive Unless mkgmap doesnt set foot = permissive to foot = yes, Garmin doesnt know what to do and will block this for all traffic including pedestrians. So we are back to the discussion about bicycles from this topic: http://www.mail-archive.com/mkgmap-dev@lists.mkgmap.org.uk/msg10590.html I would suggest to set all vehicles access with permissive to yes, so bicycle=permissive > bicycle=yes foot=permissive > foot=yes motorcar, moped, motorbike,hgv etc same except access=permissive > delete access

Am 08.08.2012 10:44, schrieb Minko:
Footways are meant to access by foot only. So no access for all other vehicles.
Unless overridden by explicit access tags. So, hw=footway + horse=yes is free for riders. hw=cycleway + motorcar=yes ( in Germay called "Fahrradstraße" ) is free for cars. hw=footway + access=yes is free for everything. Of course you can do what you want in your style file. I know, that most mappers don't fully understand the OSM access system. ;-) Chris

----- Oorspronkelijk bericht ----- Chris wrote:
hw=footway + access=yes is free for everything.
Of course you can do what you want in your style file. I know, that most mappers don't fully understand the OSM access system. ;-)
Chris
I still don't agree with you. In the potlatch editor, you see this screenshot when you are filling in the access rules. There is nothing said about access for cars on footways, so I assume the general access rule is already based on the highway type (which can be slightly different in each country). A mapper who fills in this edit form (highway=footway, general access permissive) is according to mkgmap opening this way for cars. (highway=footway, foot=yes, access=permissive, so access could not set to no) So my opinion stays, in case of access=permissive, don't set this to yes but delete the access rule otherwise you are ruining the routing.

Minko <ligfietser@online.nl> writes:
----- Oorspronkelijk bericht ----- Chris wrote:
hw=footway + access=yes is free for everything.
Of course you can do what you want in your style file. I know, that most mappers don't fully understand the OSM access system. ;-)
Chris
I still don't agree with you. In the potlatch editor, you see this screenshot when you are filling in the access rules. There is nothing said about access for cars on footways, so I assume the general access rule is already based on the highway type (which can be slightly different in each country).
The real question is what access=yes means.
A mapper who fills in this edit form (highway=footway, general access permissive) is according to mkgmap opening this way for cars. (highway=footway, foot=yes, access=permissive, so access could not set to no)
Just because potlatch does something doesn't make it right or wrong. We have to figure out semantics.
So my opinion stays, in case of access=permissive, don't set this to yes but delete the access rule otherwise you are ruining the routing.
That's adjusting mkgmap to interpret tags according to semantics that match how potlatch is, which skips the step of understanding and getting agreement on the tagging semantic. It's pretty clear from http://wiki.openstreetmap.org/wiki/Key:access that access=yes "describe[s] a general access restriction that applies to all transport modes." So highway=footway access=permissive indeed means that the way is intended principally for foot traffic but that one can use cars, horses, and bicycles. Potlatch should be setting foot=permissive instead; this probably needs to be brought up on tagging@ or talk@ because there has to be broad agreement about what tags mean among mappers, editors, and data consumers. Greg

Am 09.08.2012 19:45, schrieb Greg Troxel:
The real question is what access=yes means.
Just for curiosity I checked 4 routers, if they are routing cars over this way, tagged as hw=footway, access=yes: http://www.openstreetmap.org/browse/way/28096687 Result: OSRM : no OpenRouteService : no cloudmade : no yournavigation.org : yes So it seems that with my interpretation of the access-wiki I'm in the minority. Chris

Chris66 <chris66nrw@gmx.de> writes:
Am 09.08.2012 19:45, schrieb Greg Troxel:
The real question is what access=yes means.
Just for curiosity I checked 4 routers, if they are routing cars over this way, tagged as hw=footway, access=yes:
http://www.openstreetmap.org/browse/way/28096687
Result:
OSRM : no OpenRouteService : no cloudmade : no yournavigation.org : yes
So it seems that with my interpretation of the access-wiki I'm in the minority.
I suppose another interpretation is that to use a way, you have to have permission and you also have to have the way be suitable for your transport mode. Given that, the question remains, even if three routers avoid it.

As long as this case is still questionable, we'd better omit the access=permissive/yes etc (except access=no access=destination and access=private) tags in mkgmap until the rules are more clear. I think most mappers who tag this kind of footways are not intended to use the access tag to indicate that this road is open for all traffic, otherwise they would have set a different highway tag on them.

Hi, Am 11.08.2012 13:29, schrieb Chris66:
Am 09.08.2012 19:45, schrieb Greg Troxel:
The real question is what access=yes means.
Just for curiosity I checked 4 routers, if they are routing cars over this way, tagged as hw=footway, access=yes:
I don't understand the use of access=yes in OSM. IMHO this key allows all to use this way. Josef

Hi, Am 09.08.2012 19:38, schrieb Minko:
A mapper who fills in this edit form (highway=footway, general access permissive) is according to mkgmap opening this way for cars. (highway=footway, foot=yes, access=permissive, so access could not set to no)
And why foot=yes. Highway=footway should be enough. Regards

foot=yes is added by mkgmap for routing pedestrians on this type of road. access=no too, but since there is already a tag access=permissive, this can't be added (and thus letting cars etc pass through) So, where can we find the exact rules what comes first, access or internal highway rules?

Minko <ligfietser@online.nl> writes:
Chris wrote:
I think access=permissive is correctly mapped to access=yes in mkgmap.
I have contacted the mapper but he doesn't agree with this.
access=permissive means that the landowner gave his permission to access this footway.
Footways are meant to access by foot only. So no access for all other vehicles.
So access=permissive on highway=footway is basically an error, unless it is a walking path that isn't a public right of way, on which cars may drive and people may walk.
So the way is open for every mode of transport, cars included.
If cars are not permitted foot=permissive is the correct tagging.
If I map this road to foot=permissive, mkgmap will block this road too:
It will add access = no but it can't add foot = yes because there is already a tag foot = permissive
That's a style file bug then. In Garmin, there is no notion of permissive, so it makes sense to map permissive to yes.
Unless mkgmap doesnt set foot = permissive to foot = yes, Garmin doesnt know what to do and will block this for all traffic including pedestrians.
So we are back to the discussion about bicycles from this topic: http://www.mail-archive.com/mkgmap-dev@lists.mkgmap.org.uk/msg10590.html
I would suggest to set all vehicles access with permissive to yes, so bicycle=permissive > bicycle=yes foot=permissive > foot=yes motorcar, moped, motorbike,hgv etc same
sounds right
except access=permissive > delete access
huh? shouldn't access=foo map to for all modes bar, bar=foo before the permissive to yes (and official to yes) mappings? Is this controversial, or just a matter of getting the code to do the right thing?

I read the thread more. It seems the core of the disagreement is about whether the semantics of access belongs in java code or style files. I don't understand the problem, because it seems like Felix, who wants to have a bike map avoid bicycle=official routes, can use styles to turn those tags into bicycle=no, and then not get routed over them, because the java code will then see bicycle=no, and the fact that it would have accepted bicycle=official as bicycle=yes doesn't matter. The harder question in Felix's case is that, if I followed, he would like to not ride on a road if it has an associated official cycleway, because one can't, but if they are separate ways then maybe that's tagged bicyle=no, and that works ok.

On Wed, Aug 08, 2012 at 07:29:45AM -0400, Greg Troxel wrote:
The harder question in Felix's case is that, if I followed, he would like to not ride on a road if it has an associated official cycleway, because one can't, but if they are separate ways then maybe that's tagged bicyle=no, and that works ok.
Yes, because in Germany it is compulsory to use a cycleway where one exists next to a 'car road'. Only with highway=footway+bicycle=yes you have some choice. In Finland, cycleway use is not enforced, except maybe on highway=trunk roads. Use of cycleway is somewhat optional even according to the regulations (if short distance and it is safer to not use the cycleway). AFAIU, the Vienna convention mandates cycleway use only if it is a segregated cycleway, not a non-segregated foot/cycleway. The problem with setting bicycle=no on the 'car highway' is that it may break bicycle routing. Apparently, the Edge 705 routing is ignoring any cycleways, except those that are within about 5km of the origin or the destination. I have observed similar shortcuts in the OsmAnd offline routing. Therefore, it might be better to somehow associate the cycleway with the adjacent highway, instead of adding bicycle=no. For example, there are some sections of the highway 45 near my home tagged as bicycle=no (which is OK, because the traffic sign says so). My Garmin Edge 705 would prefer to make a 100km+ detour using some tertiary roads when routing to a destination that is less than 100km away, if the best route would be along highway 45. So far, I think that the best option is to treat yes/permissive/designated/official in the same way. That is, either preserve or delete all of these values of access=* and basically only leave the access=no|private and access=destination untouched. Marko

In Devon, at least, many paths,including DEFTA ones, are actually signed permissive. Defta permissive paths need to be closed one day a year - some v keen land owners even tell you which day!! I think most are plotted as permissive because of mapnik. -- View this message in context: http://gis.19327.n5.nabble.com/access-permissive-question-tp5719875p5720057.... Sent from the Mkgmap Development mailing list archive at Nabble.com.

Nick, Are they tagged as access=permissive and only permissive for pedestrians/walking?

On Tue, Aug 07, 2012 at 03:54:51PM +0200, Minko wrote:
In my opinion access=yes is a tag that makes no sense at all so it should be deleted anyway, and therefore access=permissive or official should be deleted instead set to yes in the internal codes.
I finally committed this (and also the removal of access=designated) in r2314. I hope it is now logical. Those who find that access=* should be preserved can simply remove or comment out the line from resources/styles/default/lines. Best regards, Marko
participants (6)
-
Chris66
-
Greg Troxel
-
Josef Latt
-
Marko Mäkelä
-
Minko
-
n Willink