Proposal:Bicycle use cycleway

Replaced by Tag:bicycle=use_sidepath

bicycle=use_cycleway
Proposal status:	Rejected (inactive)
Proposed by:	PeeWee32
Tagging:	bicycle=use_cycleway
Applies to:	highway with a classification that allows cycling generally without "bicycle forbidden sign" and with a parallel compulsory cycleway.
Definition:	This is a highway (i.e. tertiary) with a classification that allows cycling generally without "bicycle forbidden sign" and with a parallel compulsory cycleway. So this tag is only relevant for countries that have compulsory cycleways.
Statistics:
RFC start:	2013-11-12
Vote start:	2013-11-29
Vote end:	2013-12-13

What is the proposal

This tag bicycle=use_cycleway applies only to roads with a classification that allows cycling generally. When this road has a parallel compulsory cycleway (e.g. or or or ) this tag can be applied. But only when this road does NOT have a traffic sign saying it is explicitly forbidden to ride a bicycle (e.g. or ). In that case use the "bicycle=no". Legal and access implications may and will vary from country to country. No need to tag this.

Rationale

Routing

This tag will improve bicycle routing. A router can now decide to propose a route that does not use these type of roads which will be a better route for (ordinary) bicycles.

See the discussion page for details.

Rendering

A renderer can decide to render these kind of roads differently. A map can be made showing roads on which you are (not) supposed to ride with an ordinary or exceptional bicycle.

Distinguish between roads that are "forbidden" for cyclists

Here are 2 examples of types of roads on which you are not supposed to ride your bicycle.

Type1: This is a road with a bicycle forbidden sign like

Type2: This is a road with a classification that allows cycling on but because it has a parallel compusory cycleway (like in NL and DE) it is not allowed to ride your (ordinary) bicycle.

Legal authorities have made a difference between this 2 types. Why? In some countries the difference has implications on special vehicles. For example in NL a three wheel bicycle/tricycle with certain measurements is allowed to ride on the type2 road but not on type 1. In Germany there is also a legal (and access) difference between these roads.(read links) In Austria training with a race-bike is allowed on the road. (These Exceptions could be handled by the Country specific access table) In order to allow routing and rendering for different vehicles/purposes we have to distinguish between the 2 types. The question is, how?

How to distinguish between the types?

There are roughly 2 ways to distinguish.

Option A: Tag type1 with a bicycle=no AND tag type2 with a new tag such as the proposed bicycle=use_cycleway

Option B: Tag type1 AND type2 with a bicycle=no AND additionally tag Type2 with "my special vehicle=yes" in which "my special vehicle" could be every description of vehicle that is allowed to ride on type2 roads.

Option B means that we are tagging legal matters in OSM and not what we see in reality. Option B means that every time the law changes but traffic signs do not we have to change tagging. Option B also means a growth in the number of tags which does not help to keep things simple and keep overview. Option A does not have these disadvantages.

One could argue that option A (bicycle=use_cycleway) is also a legal thing but if we just agree upon the meaning of this tag it is not. bicycle=use_cycleway means: This is a road with a classification that allows cycling without a "bicycle forbidden sign" with a parallel compulsary cycleway on which you are supposed to ride your ordinary bicycle. This can be objectively seen in reality.

Examples

Picture	Tags
Cycleway: cycling is compulsory.	Cycleway: highway=cycleway (implicit:bicycle=designated) Main road: bicycle=use_cycleway
(NL) or(FR) or(DE) Cycleway: cycling is NOT compulsory.	Cycleway: highway=cycleway Possible additional tags, i.e. tags that show that the cycleway is not compulsory or mofa=* tags Main road: no explicit bicycle=* tags
Cycling is explicit not allowed	highway=* bicycle=no (explicit bicycle ban)

Some more examples in which all cycleways are compulsary cycleways.

Picture	Remarks
	NL, DE, PL: Common situation in urban and rural areas. Alternatively there could be 2 oneway cycleways. One on each side of the road.
	NL: Common situation in non-urban areas. DE, PL: Situation at some bypass- or trunk-roads.
	In this situation cycling one way you should use the cycleway. Cycling the other way you should use the main road. There are no one-way signs present on the main road so a oneway tag is not applicable.
	In this situation cycling one way you should use the cycleway. Cycling the other way you should use the main road. There are no one-way signs present on the main road so a oneway tag is not applicable.
	In this situation cycling one way you should use the cycleway. Cycling the other way you should use the main road. There are no one-way signs present on the main road so a oneway tag is not applicable.
	In this situation cycling one way you should use the cycleway. Cycling the other way you should use the main road. There are no one-way signs present on the main road so a oneway tag is not applicable.

Links

On both the German forum (in English) and Dutch forum there have been discussions on how to tag these highways. To support this tag a map is made to highlight these (and other) roads.

bicycle=use_cycleway
bicycle:forward=use_cycleway	bicycle:backward=use_cycleway

Voting

MORE TALK ABOUT THIS PROPOSAL CAN BE FOUND HERE

Please use {{vote|yes}} or {{vote|no}} and give your reasons to oppose. Use --~~~~ to sign your user name & date:

• I oppose this proposal. My compliments for the excellently well-documented proposal. I'm sorry to say that I can't support it, because it breaks backwards compatibility. We currently have the assumption within OSM that data consumers can fetch new data, without having to update rendering rules at the same time. This is also what happens in practice: most renders and routers periodically update their map data, and only rarely change their rendering or routing rules. The current proposal involves re-tagging bicycle=no into bicycle=use_cycleway for roads along an obligatory cycle path. That means that routers that used to work with old map data will suddenly stop working with new data (unless they change their rendering rules). This does not only concern the main renders and routers available on the internet, but also all forks and checkouts that people run locally. I think when we introduce new features, we should do that in such a way that it does not break existing functionality. An alternative that avoids this problem would be something like bicycle=no, bicycle:use_cycleway=yes or something along that line. I agree that the issue raised in this proposal is important, so I would be very happy to see a backwards-compatible version of this proposal. Math1985 (talk) 16:17, 29 November 2013 (UTC)

Actually it doesn't break backwards compatibility, because a bicycle=no would break routing in a lot of cases, so this has to be bicycle=yes anyway. Besides, it would be safer to handle an unknown Access-Tag as 'no', but in both cases this reasoning would be simply "against new features". --rayquaza (talk) 16:47, 29 November 2013 (UTC)

I had hoped that my comment on the tagging list made clear this tag is not breaking any backwards compatibility (except NL). --PeeWee32 20:19, 29 November 2013 (CET).

It seems it also breaks backwards compatibility in Denmark, for example here, doesn't it? https://www.openstreetmap.org/#map=19/55.68776/12.57861 Math1985 (talk) 23:38, 29 November 2013 (UTC)

Even in the Netherlands bicycle=no is mainly in use, it is simply the wrong tag! Certainly bicycle=no was used because there is no other tag and the router normally should avoid this roads. But in this case bicycle=no is tagging for router, which are not fine. --MasiMaster (talk) 00:27, 30 November 2013 (UTC)

Sorry if I missed it, but why is bicycle=no wrong in the Netherlands? Math1985 (talk) 03:18, 30 November 2013 (UTC)

I can only answer that for germany, but most of it should also apply to NL: i.e. because you may use the road if you want to get to another road at an intersection where the cycleway goes only straigth through, because several bicycles are allowed to use the road, because i.e. snow, garbage bins or cars on the cycleway allows you to use the road, because… --rayquaza (talk) 03:58, 30 November 2013 (UTC)

Math, this has to do with Dutch ( and partly German ) law. The road is not forbidden for bicycles, but it is mandatory to use the cycling lane. The latter only when the cycling lane is clearly next to and belonging the road, and also dependent on the type of bicycle. There are other types of road where cycling is forbidden by law, governed by roadtype ( freeways etc ) or by a specific sign. These are correctly tagged with bicycle=no. What this proposal does is offer a solution for those situations mainly in NL en BD where routers should send cyclist on a lane, without interfering with legal tagging. So it is an addition to, not a replacement of bicycle=no. Noordfiets (talk) 13:06, 30 November 2013 (UTC) fixed indention--rayquaza (talk) 16:48, 30 November 2013 (UTC)

The German law says: "Der Radverkehr darf nicht die Fahrbahn, sondern muss den Radweg benutzen (Radwegbenutzungspflicht)". The Dutch law is less explicit, but the idea is similar. I am aware of the exceptions for oversize bicycles, but I would guess they account for less than 1/1000 of all bicycles. I cannot find an exception for obstacles in either the Dutch or German law, which article would that be? In any case, obstacles cannot be taken into account for routing anyway. Would you be able to give me a link to an intersection where cyclists must use the main road? —Preceding unsigned comment added by Math1985 (talk • contribs) 2013-11-30T16:03:17Z

Sure. If you're approaching from south and want to leave to west you're free to not use the comulsary cycleway. --rayquaza (talk) 16:48, 30 November 2013 (UTC)

If you zoom in on Google Maps, you see that there is in fact a short connection between the Eastern cycling path and the main road. I think that the current mapping is wrong, and that that connection should be mapped as highway=path or highway=cycleway (or in this case even highway=service because it's part of a driveway). In any case, I believe it would be illegal here to use the main road of the Frankfurter Strasse as a cyclist, even when turning into Kortenbacher Weg. Therefore, routing over the main road would be incorrect. I wouldn't be surprised if situations like you describe indeed exist, but this isn't one, in my opinion. Math1985 (talk) 20:18, 30 November 2013 (UTC)

Oh, erm… Ok, it was such a situation some time ago, apparently that has changed. Though it might be still allowed to use the road for a normal turn instead of a hook turn, that's not a "must use the road" as you asked for. --rayquaza (talk) 21:03, 30 November 2013 (UTC)

Example for no connection to the secondary. --MasiMaster (talk) 15:40, 1 December 2013 (UTC)

Thank you for that example. I'm still not sure what the law would be in such situations. Wouldn't it be correct to ride through the grass? It's hard to tell from the Google Maps imagery, but the grass seems darker there, so that might be what actually people do. —Preceding unsigned comment added by Math1985 (talk • contribs) 2013-12-03T20:29:00Z

Another question: some countries, like the UK, do not have the concept of bike paths that are obligatory to use. Does that mean that in such countries, this tag should not be used? Math1985 (talk) 03:40, 30 November 2013 (UTC)

That's right. --rayquaza (talk) 03:58, 30 November 2013 (UTC)

Yes: See definition. --PeeWee32 (talk) 08:10, 30 November 2013 (CET)

• I approve this proposal. I don't think it's perfect (How could a router ever decide this? What about avoiding compulsary cycleways, but using normal ones?), but it's definitly better than the current situation. --rayquaza (talk) 16:47, 29 November 2013 (UTC)

A router can give an option, if he may route over these and other roads/cycleways. For getting to a house which are not touch the cycleway, the priority of the use_cycleway-road could be 50 or 90% lower, so the destination is attainably. With bicycle=no there is no connection to the destination. --MasiMaster (talk) 00:27, 30 November 2013 (UTC)

I think you didn't get my question. But you don't need to convince me, so let's discuss this after the proposal. --rayquaza (talk) 03:58, 30 November 2013 (UTC)

• I oppose this proposal. I don't believe that the problem that this proposal is trying to fix actually exists. A "law obiding" router will prefer cycleways over normal roads and the forward/backward case is already covered by using/not using oneway tags on the cycleway. A non-"law obiding" router will ignore the tagging in any case. Naturally tagging the "main" road with bicycle=no should simply not be done in the relevant cases. SimonPoole (talk) 18:39, 29 November 2013 (UTC)

Simon, I certainly think that a problem exists, because today there is no clear way to tag whether a cyclist can use the road or has to use a compulsory cycleway. It is not only about routing, it is about mapping what is on the ground. Nillerdk (talk) 19:36, 29 November 2013 (UTC)

The problem that exists, is that the roads are being tagged wrongly with bicycle=no in a situation where it is not a hard no (for example if you want to visit a house which is on the road and it can't be reached via the cycleway etc, the arguments have all been made so I wont repeat them all here again. It is not an attribute of the road that you are not supposed to ride on it, it is a result of the cycleway being there, since this is already in the database, no further information is needed. SimonPoole (talk) 21:06, 29 November 2013 (UTC)

Simon, regarding your last sentence I must say you are right, in theory that is. If compulsory cycleways would be tagged properly (that’s an other issue to be discussed later) a router could e.g. give the main road a much lower priority for bicycles then a normal road without adjacent compulsory cycleway. But the router has to do that based on location / geometries of both cycleway and road. A difficult and imho almost impossible task . Bicycle=use_cycleway is a tag to make this task a lot easier. If I exaggerate your approach a little more I could say that a traffic sign “bicycle forbidden” could be tagged as a node next to the road with additionall tag telling in which direction the sign is facing. This should give the router enough information. In OSM we do not do it this way. We add the attribute to the road. This makes it a lot easier to handle for routers. PeeWee32 (talk) 18:25, 30 November 2013 (UTC)

Simon. In NL these roads are tagged with bicycle=no and in most other countries they are not (because there is no explicit "bicycle forbidden" sign). A problem to be solved. In NL some odd vehicles may ride these roads but currently OSM is not able to tell us where this is possible. Another issue is that access info for these roads depends on the status of the parallel cycleway. If it’s a compulsory cycleway it is a different situation then a non compulsory cycleway. I can imagine a bicycle router taking the main road when the parallel cycleway is not compulsary. --PeeWee32 21:04, 29 November 2013 (CET).

Simon, you are right, this tag will not solve all tagging problems. It is only a part for a nice cycleway-tagging. Maybe other proposal will follow to distinguish compulsory cycleways from cycleways without compulsory by a tag on the cycleway. Note that "law obiding" router will only prefer cycleways, if these are compulsory. But there are also other cycleway without compulsory. I.e. in Germany there are cyclists which hate most cycleways, because it is to dangerous to cycle there. They avoid most non-compulsory cycleways legally! Further if the road is untagged (i understand that you prefer this), the router have to know the relation between the cycleway and the road (or prefer cycleway much more times instead all roads-types). With this new tag it is possible to avoid roads(+the cycleway) with compulsory cycleways, i.e. if you think cycleways are more dangerous. --MasiMaster (talk) 00:27, 30 November 2013 (UTC)

Why should bicycle=no not be used for the main road, in the Netherlands or Germany? Math1985 (talk) 03:38, 30 November 2013 (UTC)

Answered under the first vote --rayquaza (talk) 03:58, 30 November 2013 (UTC)

In NL we have 1) roads accessable for bicycles (no tag or bicycle=yes) , 2) roads with explicit ban (bicycle forbidden sign, tagged bicycle=no) and 3) roads that are more or less forbidden for ordinary bicycles unless (with a lot of exceptions) without a "bicycle forbidden sign". It's either one of the 3. A bicycle=no applies to number 2. This proposal is about the number 3 category and is only relevant for countries that have compulsory cycleways.PeeWee32 07:22, 30 November 2013 (CET).

I'm still not convinced that we cannot use bicycle=no for 2 and 3, and use subtags to distinguish them. Math1985 (talk) 16:06, 30 November 2013 (UTC)

Because that would break current routing: In germany it is often not possible to do a left-turn from the (still compulsary!) cycleway, which means that you're allowed to leave it for this turn. --rayquaza (talk) 16:48, 30 November 2013 (UTC)

• I approve this proposal. This is a well thought out proposal which I would like to use myself. Nillerdk (talk) 19:36, 29 November 2013 (UTC)

I abstain. I must admit I don't see what the problem is being solved. If there are parallel ways, one for cyclists and one for motor vehicles then surely it is implicit that cycles should follow the cycleway and if the local laws state that if a cycleway is present then cycles can't use other roads then surely just adding a bicycle=no solves it, that maps what is on the ground. Routers need to be aware of parallel ways and the type of vehicle they are routing. Bigfatfrog67 (talk) 20:14, 29 November 2013 (UTC)

It is not implicit that cyclists have to use all cycleway! There are many reasons to use the road: much more save, faster, often the cycleway is blocked by parking cars (i.e. in Germany), for left turn on crossings the cycleway goes over 2 or sometimes 3 or more traffic lights while the cars waiting only at 1. --MasiMaster (talk) 00:27, 30 November 2013 (UTC)

According to the German law, the round blue sign means: "Der Radverkehr darf nicht die Fahrbahn, sondern muss den Radweg benutzen." (Translation: cycling traffic can not use the main road, but must use the cycling road"). So in Germany (and in the Netherlands, but not in the UK), it would be illegal to use the main road because its safer or faster. Math1985 (talk) 03:53, 30 November 2013 (UTC)

You should re-read the current law, not a long outdated one ;-) --rayquaza (talk) 03:58, 30 November 2013 (UTC)

My information comes from the Bundesministerium des Justiz. Is there a newer version of the law somewhere available? —Preceding unsigned comment added by Math1985 (talk • contribs) 2013-11-30T16:03:17Z

That's the current one. Sorry for the aggressive tone, I thought you meant another sentence. But I don't know where exactly this is written there and can't find it atm… --rayquaza (talk) 16:48, 30 November 2013 (UTC)

It's right that "cycling traffic can not use the main road, but must use the cycling road". But there are some other laws, 2 german examples: Voraussetzung für die Kennzeichnung ist, daß die Benutzung des Radweges nach der Beschaffenheit und dem Zustand zumutbar / Die Führer anderer Fahrräder sollen in der Regel dann, wenn die Benutzung des Radweges nach den Umständen des Einzelfalles unzumutbar ist, nicht beanstandet werden, wenn sie den Radweg nicht benutzen. (translation: Prerequisite to set the compulsory-signs is that the cycleway is suitable. / Other (not "normal") bicycles should not be ticked, if the cycleway is not practicable.) Source: Zu § 2, Leftnumber 15, 16 & 23 But it is often not suitable. Another, § 27: Mehr als 15 Rad Fahrende dürfen einen geschlossenen Verband bilden. Dann dürfen sie zu zweit nebeneinander auf der Fahrbahn fahren. (translation: more than 15 cyclists can cycle on the road, 2 in a row).

Anyway, different rights in different countries. With this tag (a softer no than access=no or private, and different) it could be possible to route following the country-specific rules. --MasiMaster (talk) 00:20, 2 December 2013 (UTC)

I would suggest that because of all the discussion here this proposal needs to go back a step and thrash this all out again before a voye. Vote's shouldn't be this contentious. Bigfatfrog67 (talk) 19:34, 1 December 2013 (UTC)

• I approve this proposal.--Dieterdreist (talk) 23:54, 29 November 2013 (UTC)

• I oppose this proposal. I completely disagree why tagging the main road with bicycle=no is wrong and called "mapping for the renderer" only because there are a few exceptions in some cases you might have to cycle on the main road. Cycleways blocked with cars, snow etc is not a valid reason to claim that bicycle=no is invalid and to introduce another bicycle=* tag like bicycle=use_cycleway only makes it more confusing for the renderers. I agree with Math1985 that we should work out a better backwards-compatible version of this proposal because I also understand that are differences in bicycle=no. Why not tag the main road with a tag cycleway=compulsory or something alike (in the mailing list bicycle:restriction=use_cycleway has been mentioned which is fine)? Instead of bicycle=no use cycleway=compulsory or bicycle:restriction=use_cycleway, which might be used in accordance with bicycle=no to solve the problem of backwards compatibility.--ligfietser (talk) 08:50, 30 November 2013 (UTC)
  • I dislike "cycleway=compulsory" because it is not clear that cycleway it is not clear that it is not about bike lane, but "bicycle:restriction=use_cycleway" (or maybe "bicycle:restriction:use_cycleway=yes", there may be some other kind of restriction frequently appearing in say Asia or Africa) may be better than what is proposed here Bulwersator (talk) 10:23, 30 November 2013 (UTC)

Imagine you ride a wide bicycle for transportation. It is not suitable/allowed to use the cycleway. Currently you have to say to the router: avoid cycleways, and allow roads with compulsory cycleways, (bicycle=no). But whats about with the roads that strictly forbidden for all bicycles? These roads will available too in this example. Other Example: In Germany it is allowed to use the road, if you are in a group with 16 cyclists (with normal bikes), even if there is a compulsory cycleway. We have to have a different tagging for different signs.

Backward-compatible is an argument! But first bicycle=no is the current tagging only in NL, for other countries its only a new tag on the road. Second, we have in the access-tag group a few similar tags: no, private, destination, permissive, which all say, "no access, but with some exceptions". We don't have access=no + private=yes. So a bicycle=* tag fit well in our scheme. --MasiMaster (talk) 16:20, 30 November 2013 (UTC)

• I oppose this proposal. We should tag what we see. This proposal suggests to set no tags when there's a "bicycle free" sign, and to set a tag when there is no sign. This is just the wrong way around. What if a mapper comes across a "bicycle free" sign in situ. He will not set any tag, because your proposal forbids that. The next day another mapper sees that the last change date was before this proposal was approved. So he thinks that the new tag is just missing and adds it with good intentions, because of the huge probability that the tag belongs there. We'll end up in a mess with the tag still missing on most roads, while incorrectly added to others. The correct approach is to set bicycle=yes on the road if there's a "bicycle free" sign, and to omit the tag if that sign is not there. Then it's up to the routing engine to derive from distances and national laws whether a road is forbidden for cyclists. This requires some "fuzzy logic", because the laws are fuzzy too. The laws don't define the maximum road-cycleway distance up to which the cycleway is compulsory. So that's a tough task for routers. But it would even more be a tough task for mappers. We just cannot decide if cyclists may use a road 20m beneath a cycleway. We NEED to leave it over to the applications, just like the interpretation of walking speed etc. --Fkv (talk) 10:01, 30 November 2013 (UTC)
  • "We should tag what we see." - and we tag here that we see cycleway along the road. Note also that relying on lack of tag is a bad idea, most of newly added roads are not fully tagged. Bulwersator (talk) 10:20, 30 November 2013 (UTC)
  • "We should tag what we see." - I agree. This proposal is about tagging what we see. If you see a bicycle forbidden sign: tag bicycle=no. If you don't but you do see a compulsary cycleway (indicated by a clearly visable traffic sign) tag bicycle=use_cycleway PeeWee32 (talk) 12:40, 30 November 2013 (CET)

• I abstain from this proposal. In my opinion this is a completely unnecessary tag. As already have been said this proposal asks the mapper to tag something that is not there (no sign -> tag ; sign -> no tag) - that doesn't make any sense at all. But if I have to choose between the completely non-sense bicycle=no tagging that some people demand and this useless tag, I would clearly choose this tag. --Imagic (talk) 10:48, 30 November 2013 (UTC)
  • I think this proposal is about tagging only what we see. See my remark a few lines up. If you have any suggestion on how to improve bicycle routing on the type of roads we are talking about please do so. PeeWee32 (talk) 14:13, 30 November 2013 (UTC)
  • If you use such strong words as 'completely nonsense', you could at least explain why you are of that opinion. Perhaps it would be a start to indicate in which country you believe that bicycle=no is complete nonsense. Math1985 (talk) 16:06, 30 November 2013 (UTC)

Fkv and Imagic, I don't understand your example: "This proposal suggests to set no tags when there's a 'bicycle free' sign, and to set a tag when there is no sign." We propose not a tagging in relation with a 'bicycle free' sign. The proposal is only for compulsory cycleways. A 'bicycle free'-sign correlate to bicycle=yes, which is untouched by this proposal. The 'compulsory cycleway'-sign relate to the cycleway and the road, so I think we had to tag both (or add a relation between cycleway & road). One part of the result of this tagging is also, to avoid all roads(+their cycleways) with compulsory cycleway, which are currently not possible. --MasiMaster (talk) 16:40, 30 November 2013 (UTC)

• I abstain from this proposal. - for no I am not truly convinced that it is a good idea. But backwards compatibility is not something that never should be broken and this tags existing feature (compulsory cycleway). But for now I think that we should avoid adding additional values to this key, I think that special key ("bicycle:restriction:use_cycleway=yes") for special situation is a better idea. Especially as there is possibility of other special case that may exist together with this situation Bulwersator (talk) 20:34, 30 November 2013 (UTC)
  • Can we get example of routing software or just an example of situation where it is impossible to currently produce correct solution and this tag would make this possible? Bulwersator (talk) 10:04, 3 December 2013 (UTC)
• I oppose this proposal. - This is a proposal looking for a problem that largely could be worked out with already existing tags (bicycle=no, bicycle=destination) and mapping for completeness. Paul Johnson (talk) 21:48, 30 November 2013 (UTC)
  • "destination Only when traveling to this element, e.g. customer parking lots." (from Key:access) so this would change meaning of existing value, what is even worse. Bulwersator (talk) 09:24, 1 December 2013 (UTC)
• I oppose this proposal. - The access rights implied by the proposed are not sufficiently well-defined for this to be used on its own as a primary access-tag value. If a solution is needed to mark such roads in between the extremes of bicycle=yes and bicycle=no (possibly combined with more specific over-riding tags for specific sub-types of bike, as in Option B above) then I would advocate an alternative approach, as follows. In the bicycle key, we introduce a more general-purpose access tag of bicycle=restricted, and combine this with a second tag that provides a machine-readable value for the precise restriction imposed along the lines of bicycle:restriction=DE:use_cycleway. This way, routers without specific knowledge can sensibly interpret bicycle=restricted as "you probably can't cycle here, but there may be some exceptions" and warn users if a route needs to include such a section. On the other hand routers that do know about the specific restrictions (which will no doubt vary from country to country, and may also include local variations) have a second key that they can check to learn about the precise details. The bicycle:restriction key values such as DE:use_cycleway would be specifically tied to a well-defined legal restriction, and each restriction can have an associated list of implied tags (eg tricycle=yes, bicycle_with_trailer=yes, bicycle=destination, etc.) which could be used by routers without mappers needing to specify them explicitly on each way. -- Rjw62 (talk) 23:02, 30 November 2013 (UTC)

• I approve this proposal. It is a small Step towards a better bicycle routing. The properties of cycleways are very often different from those of a road e.g. oneway=*. The resulting highway=path and highway=cycleway should be indicated on the street, to make a safe routing possible. That cycleways in some or many countries are usually oneway roads is also not indicated by a sign, similar to bicycle=use_cycleway.--Cracklinrain (talk) 00:49, 1 December 2013 (UTC)
• I approve this proposal. We have this same problem in Belgium, with equal exceptions when cyclists are still allowed to use the main road. bicycle=no/destination are wrong, so a tag like this would really be welcome. This tag can also be of great help for routers that now can only guess if a cycleway should be used or not. --Eimai (talk) 10:34, 1 December 2013 (UTC)

• I approve this proposal. Much needed value. Using bicycle=no would be a serious offense to the definition of =no, meaning its definition would become overloaded, and bicycle=destination likewise. The only alternative I can think of which could eventually improve the same situations would be street area or relation but that's not very realistic to happen even in the far future on large scale. --Ij (talk) 10:53, 1 December 2013 (UTC)

• I oppose this proposal. This is actually more a routing problem than a tagging problem. Good bicycle routing software will always prefer cycleways along the road anyway, so having a bicycle=use_cycleway tag shouldn't make any practical difference in most situations (in situations where it does, the cycleway probably isn't correctly interconnected to the main road at all relevant intersections, otherwise the router would have considered it). Furthermore, exceptions to this rule (like e.g. for racing cycles, tricycles etc.) can also easily be handled by the router by simply preferring the road and not the cycleway in these situations. So, just tag the road with neither bicycle=no nor bicycle=use_cycleway, and the router will be able to do its job. Thus, no need for such a tag. --Emkey08 (talk) 12:54, 1 December 2013 (UTC)

No, a good bicycle router will ask the user, if (s)he wants to use (non-compulsory) cycleways or not. I'm not sure in which country do you know cycleways... In Germany it is much more dangerous to ride on cycleways instead on the road. It's a fact! So we should make a difference between non-compulsory and compulsory cycleways to get a choice. This proposal is only one step to a better bicycle tagging. Half agree to "better no tag instead of bicycle=no". --MasiMaster (talk) 15:35, 1 December 2013 (UTC)

This means bicycle=use_cycleway would only improve routing for those users who prefer to cycle on the road unless there is a compulsory cycleway. In any other case the tag provides no new information. Neither the routing for users who always want to cycle on cycleways will be changed (as the router will always prefer the cycleway anyway in this case), nor the routing for users who always want to cycle on the road will be changed (as the router will never prefer the cycleway anyway in this case). --Emkey08 (talk) 16:39, 1 December 2013 (UTC)

Not only. Some router switch sometimes between cycleway and road, if the cycleway is a bit longer. So you have to set the cycleway priority higher. The negative result is, that a route will choose which could be longer by distance and time (i.e. a unclassified/tertiary is a shorter way). Maybe this is the reason why the Netherlands set bicycle=no to the road. The use_cycleway tag could be used to set this roads a lower priority, so the router will primary choose the parallel cycleway. (I think unclassified/tertiary is in most cases better for cycling than cycleway.?) --MasiMaster (talk) 17:56, 1 December 2013 (UTC)

Yes, but this exact problem could be solved *much* easier by simply adapting the way weights within the routing software. As an example, for users who prefer cycleways, apply an internal factor of 1.1 (or even less - whatever factor is reasonable) to all road weights. For users who prefer roads, apply the same factor to all cycleways. The router will now use your preferred way type unless it's more than 10 % longer, without the need of any additional tags. Hence my original reasoning of arguing that this is "more a routing problem than a tagging problem". --Emkey08 (talk) 19:58, 1 December 2013 (UTC)

• I approve this proposal. This is a good step to better bicycle routing. In Finland the road rules say that one can use main road despite compulsory cycle way if the cycle way is on the left side and you ride on main road only for a short distance. Also, principle where there is tag only when there is sign fails anyway with bicycle road rules because signs on the other road that is cycleway or sidewalk affect road rules of the vehicle lanes of the road. Sometimes compulsory cycleway can be routed quite faraway from the main road in the countryside, and then it might be problematic for an algorithm to decide if the nearby cycleway is related to a road. --Thv (talk) 13:17, 1 December 2013 (UTC)

Can you explain how you think the tag would improve bicycle routing? Since we're seeing many exceptions for compulsory cycle ways which depend on local law (like the short distance ride you mentioned, and many others in other countries), and because of potential cycleway interconnection difficulties at intersections, a router can in general only interpret a bicycle=use_cycleway tag as an advice to use the cycleway. This however provides no new information to the router, as it would have used the cycleway for cycle routing in either case. I thus strongly doubt that this tag can actually improve bicycle routing in any way. --Emkey08 (talk) 15:02, 1 December 2013 (UTC)

• I approve this proposal. I like the proposal --chris66 (talk) 10:28, 2 December 2013 (UTC)

• I oppose this proposal. Well formed proposal, but I doubt if tags in OSM should be introduced to solve problems in routingsoftware for a particular country. OSM should reflect the legal status of a road. Routing software should apply a mechanism to give weigth to a cycleway depending on distance. This than does reflect the legal situation in NL where a cycleway that is not directly adjecent to a road is not mandatory by default. Furthermore it seems that the proposal is mainly focussed on an adaptation of the Garmin routingsoftware that cannot cope with this situation. I also doubt if it's really a big problem: if a cycleway is adjecent to a road and the routingsoftware draws a line on the road surely it is my responsibillaty as a cyclist to use the cycleway? Noordfiets (talk) 10:01, 2 December 2013 (UTC)

This is largely my assessment as well. While the situation is by no means unique (states wishing to receive certain federal highway dollars in the US operate in the same fashion, meaning pretty much everywhere but Kansas), this whole situation sounds like a hack to deal with bad routing software. Existing access tags and mode-specific turn relations (yes, I realize there would be many) already exist to handle this situation, and should translate to existing routing engines. Paul Johnson (talk) 17:43, 2 December 2013 (UTC)

Though I wish it were true that routing software would solve the problem, I (reluctantly) believe that this tag is currently the best option to resolve differences in tagging between e.g. the Netherlands and Germany and ensure that universal routing algorithms will make the correct choice in the place I live and map (viz. The Netherlands). The alternative of putting turn restrictions on all the side connections along a compulsory cycleway to prevent turns onto the road is more work and more likely to lead to mistakes as mappers add driveways (as the map becomes more detailed). Frankl2009 (talk) 10:30, 4 December 2013 (UTC)

@ Noordfiets. As far as routing is concerned I am afraid I can not convince you but let's just look at the access part of the proposal. You state "OSM should reflect the legal status of a road". As a Dutchman you are well aware that the legal status of "bicycle=use_cycleway" road is different from a road with a "bicycle forbidden" sign and from an ordinary road that does not have a parallel compulsory cycleway. Why not map this? This mapreflects more or less the legal status. Ofcourse this is only possible if we map this. Does this convince you? PeeWee32 (talk) 04:35, 6 December 2013 (UTC)

• I approve this proposal. A reluctant vote yes, because I am not convinced that routing algorithms will be able to get it right for the place I live and I prefer hints to ensure accurate routing. Frankl2009 (talk) 10:30, 4 December 2013 (UTC)

@Frankl2009. As you are well aware in NL the bicycle=use_cycleway is very close to a bicycle=no for ordinary bikes. Both Openfietsmap (garmin) and Osmand (android) will never route you over these bicyce=use_cycleway roads since they are all (wrongly) tagged with bicycle=no. The only thing these routers have to do in NL is considder these tag the same for ordinary bicycles. Should not be a problem. --PeeWee32 (talk) 16:30, 6 December 2013 (UTC)

• I approve this proposal. I like the proposal --AmiciGiuseppe_IT 19:49, 4 December 2013 (UTC)

• I oppose this proposal. is not prioritary, but if you like to have a tag just for countries with this kind of law, maybe you should find a different tag. not so usefull btw as routing of navigation software generally suggest automatically to prefer cycleways when you set bicycle as vehicle. --Giardia (talk) 00:09, 5 December 2013 (UTC)

Do you have compulsary cycleways in Italy? If yes.. what are the access implications on the parallel road for bicyles? If no.. what would you suggest we use for tagging on parallel roads since this has implications on access for bicycles? --PeeWee32 (talk) 16:09, 6 December 2013 (UTC)

• I oppose this proposal. like Giardia --Madeco (talk) 13:00, 6 December 2013 (UTC)

I have the same question I asked Giardia for you. If you could answer I would very much apreciate. --PeeWee32 (talk) 10:40, 7 December 2013 (UTC)

• I oppose this proposal. I agree with Noordfiets here. This isn't really a tagging problem but a routing engine problem which should always give preference to parallel cycleways. Furthermore a restriction tag should be a direct explicit property of the highway on which it is applied and not implicit due of the presence of another highway. I also doubt that such a tag will be correctly used by mappers, either because they don't understand it or because they map from imagery where you can't see any road signs. --polderrunner (talk) 22:26, 12 December 2013 (UTC)

For the record. A router should not always give preference to cycleways. Not when you choose SHORTEST route. See the picture at the bottom of the talk site. Or look at this example where the router takes the shortest route and has no way of knowing it should not takes these roads (because there is no bicycle access tagged).--PeeWee32 (talk) 15:50, 13 December 2013 (UTC)