From OpenStreetMap Wiki
Jump to: navigation, search

More usage instructions

This paragraph is linked to by the specification page as a supplement. It deals with the way to tag noexit=yes according to that page, not according to other wishes and opinions (including of this discussion page). Do not comment it — just correct it — inside but in the next paragraph.

The best way to put an end to the various misunderstandings about noexit=yes is to not use it at all and simply add in sufficient detail to explain why there's no exit - like add in a fence, wall or other obstruction. rather tha say that, according to a careful reading of the terms of its wiki page, it's only made to clear a validation warning and it applies almost ONLY TO THIS SITUATION:


  • Open JOSM;
  • File>New Layer;
  • draw two ways tagged with highway=primary;
  • One way must be perpendicular to the other and its end node must be very near to the other way.
  • select both ways
  • Click Windows>Validation Results>Validation
  • You get Warning: Way end node near other highway -> ways & node rounded up by Validation Layer
  • Now add noexit=yes to the node and
  • do Windows>Validation Results>Validation again.
  • The Warning disappeared.

That's the meaning of noexit=yes: nothing was changed to the meaning the map data and to its other tags, no traffic routing was changed, noexit=yes just disabled the Warning by JOSM and other validators.

That's the only meaning of noexit=yes: disable warnings by telling that the configuration is correct as follows.

  • ======O ---------- : noexit=yes on the node to indicate that the O - gap, which is the reason why one cannot exit from that end of =====, is perfectly real and normal and that quality assurance software should NOT warn about it and that no one should try to repair what seems an "error".
  • There may be similar situations like:
  • ======O       O====== with the same names on the two parts, because the missing part crumbled down, which is a very good explanation to put in a note=* of both noexit=yes tags.

In short, noexit=yes must never be used to indicate that one cannot pass, it's only used to indicate that this situation is normal as indicated by other tags or road layout. It's totally ignored by GPS routers. noexit does not indicate to which means of transportation it would apply noexit does not indicate in which direction it would apply put on a way, noexit does not indicate at what end it would apply

In particular, noexit=yes must not be used:

  • ======O  : it's obvious that nothing can pass, no warning, noexit is unnecessary.
  • ======O----------  : if cars cannot pass to path ==>>--, noexit=yes is superfluous: it's wrong for pedestrians and the status for cars is made obvious on ------- with motor_vehicle=no, no reason for any warning (to disable)
  • ====== : on a way, because, even if a valid noexit=yes condition is required, one must obviously indicate at which end node it's required

This traffic sign is not to be tagged with noexit=yes because it is not placed where the restriction occurs (it exists because a driving driver does not see the other end of the road, but that's not true for someone looking a a map). {tag|noexit||yes}} may be necessary on the other end, but depends on what's going on there as just explaind here.

More usage instructions: comments about

I wish that the guy who invented noexit=yes had thought twice, foreseen all this and called it no_warning

Noexit=yes is useful

When Noexit=yes is rendered on a map, it is a very practical and helpful indication of where there is a real cul-de-sac. The current discussion seems to completely ignore the needs of the common map-user who has no access to the tags. He just has his map. An internal, free-form comment in a note=* tag is only visible to mappers but is completely useless to map users and cannot be evaluated by renderers as the note tag may contain anything in any form.

Here's a practical example: Imagine you are planning a hiking tour and you don't want to use the main paved roads. Look at this area: Ignoring noexits. There is several trails there that just end. Now tell me which of those are real cul-de sacs and which are just mapped incompletely? Now look at this map of the same area Showing Noexits. Now it is immediately clear which ways you don't need to try because they are confirmed dead ends. There has been the reasoning "You can only use a map when it is complete, cou can never use ways that seem to end". But this is unrealistic. There is no such thing as a complete map when trails are concerned, if you don't want to use the OSM map you have to use the official topological maps - and those are known to be incomplete, inaccurate and are only renewed every 10 years. You always have to take chances when using unknown trails on a map - and noexit=yes help you to avoid confirmed dead ends when doing this. --Nop 22:03, 22 November 2009 (UTC)

What you say here can be said for many other attributs. For example, the oneway tag. When a road does not carry the oneway tag, we assume that the road is bidirectional. It's the same if the road ends up no where. If the mapping is not finished, I can understand that you don't trust the information. But we have to assume that when a region is well mapped in details, a road ending without the tag noexit=yes IS a cul-de-sac. If we don't do that, each segment of highway will carry ten, twenty, hundreds tags to be sure that nothing is implied. And my description does not say that noexit=yes has to be removed, it is surely useful in some cases. But I see this tag in urban areas where all streets are 110% mapped in high detail and all cul-de sac, even the obvious ones, are tagged with noexit=yes. -- Pieren 11:43, 23 November 2009 (UTC)
I agree that the tag is of little importance when used on roads in a well-mapped environment. But it has a lot of importance when used on tracks and trails in a rural environment. But the rule in OSM is: If a tag is superfluous to you and useful to others - just ignore it, don't interfere with it. --Nop 12:40, 23 November 2009 (UTC)
It does not matter to map users if it is a unfinished mapping or a real dead end. It is plain stupid to plan a trip where you expect going through such a "ending road" anyway. If it is simply a unfinished mapping matter, you still don't know how the road might turn and where it might take you. (And "if" we have problems with unfinished mapping, there might as well be unfinished noexists.) --Henriko 23:28, 7 March 2010 (UTC)
Sometimes (really, almost always) OSM users are map makers more than end users. For example node Node 298175190 (XML, iD, JOSM, Potlatch2, history), the end of a tram stop platform in between the road carriageways in separated directions would be, to a quality control tool, indistinguishable from the other end of that platform (the end which has the pedestrian crossing), without the noexit=yes. In an ideal world there could be areas with, say, landuse=divider and ways with barrier=curb etc. drawn to convey the same information, but an interim solution is to add the tag when there's no way forward for any mode of transport. (IMO, if it's a regular cul-de-sac and there's anything already drawn (a building, a park, river, ...) then the other side of that feature is far enough that any "way ends close to another highway but isn't connected" check shouldn't be fooled. Alv
putting a noexit=yes onto the end node is like saying I'm going to edit the map because I can't be bothered to edit the map correctly. Maybe it should end with a fixme ? --pmailkeey 2016:4:10

Candidate for the most stupid tag competition?

IF the way stops, then it is a cul-de-sac. That's it ! No tag required. If it continues but is not finished/surveyed, then add a fixme=* or note=* but please, not noexit=no. -- Pieren 23:36, 19 November 2009 (UTC)

For regular streets I agree, there's either some natural feature blocking the route onwards, or a footway/cycleway extending from the road endpoint, both of which already make it quite clear that the road does end. Yet, if this node wasn't tagged with a noexit=yes no, how could consistency check software know that it isn't and shouldn't be connected to any of the ways within few meters (as in the "way ends near other highway" check as implemented in various tools)? The only connection is at the other end of the tram platform. Alv 07:27, 20 November 2009 (UTC)
noexit=no doesn't make any sense as it is the default of any continuous road. noexit=yes is useful and is being used and rendered as already discussed twice on this page. --Nop 09:21, 20 November 2009 (UTC)
Sure, missed/miswrote that one, meant noexit=yes, which the first comment could have been misunderstood to be talking about, too. Alv 10:43, 20 November 2009 (UTC)
noexit=yes is useful only in some cases as Alv said. But noexit=no is ridiculous. Even if one person decided to add it in osmarender, it is ridiculous. --Pieren 16:34, 22 November 2009 (UTC)
Despite the agressive title ;-) I'm all for removing that tag wich causes (IMHO) more troubles than the few "consistency checks" it helps to resolve. Or at least should it be made clear and be moved to some "consistency help tag checks" sletuffe 16:50, 22 November 2009 (UTC)
The tag is not just a consistency helper. It does carry useful information when rendered on a map. --Nop 17:59, 22 November 2009 (UTC)
How? Consider one street that ends because it hasn't actually been built any longer, tagged noexit=yes at the end, and then another street that ends just because some mapper got bored half way through, not tagged with noexit. What possible useful thing could some renderer do different at the way that perhaps continues, but the renderer have no clue where? Perhaps the street actually ends 400 meter later, in real life, just like the other one. --Henriko (talk) 21:16, 8 August 2017 (UTC)
I wasn't very clear with my "I'm all for removing that tag". This tag was used, and is still used, so there is no reasons to blank the page or delete it from the osm database. (And I just discovered there war an edit war going on). What I was suggesting was to add a footnote on the mail page to discourage it's use for future mappers, but in no way change the description. Pieren's edit were not good. The proposition down there to add a text saying "noexit=no should be avoided in favor of fixme=continue if you intend to say that a road continues and was not mapped." looks a good first step to me. However, I would go even further by saying "noexit=yes is a tag to help renderers/data checkers to guess it's a cul-de-sac, even if that information could be derived from the topology, so feel free to use it if you wish but it is not needed that much". But maybe I am wrong, and cases exists where topology cannot give it for sure sletuffe 19:20, 22 November 2009 (UTC)
If there's a cul-de-sac, put in a highway=turning_circle or similar; that should be (and I hope is) enough to shut up the validator. However, sometimes there isn't, i.e. the road really does have a dead end, and noexit=yes is a reasonable solution. StephenTX (talk) 20:01, 21 November 2016 (UTC)
It's an excellent candidate for
  • The most stupid tag
  • The most destructive tag (as it encourages abortive data)

If a validator flags up a 'node not connected to a way error' then for Pete's sake fix it properly, not with a cop-out tag like this one! Alternatively, fix the validator! After all, the warning is a validator's 'fixme' signal - so there's something wrong with the map - i.e. the validator's misreading it --pmailkeey 2016:4:10

Sometimes you really do have one road end (or two both end) really close enough to another. For instance, I had a case where it appears there was a driveway on private property and the property was subdivided, and one of the new owners put a fence across the driveway with no gate, making it impassable. It's still one strip of concrete, but I mapped it as two ways with end nodes nearly touching. The only way to tell the validator it's a false positive is with noexit=yes. Turning down the sensitivity would miss true positives where the ends are much further apart. StephenTX (talk) 20:01, 21 November 2016 (UTC)

noexit new proposition

  • noexit=yes To label a cul-de-sac or dead end street
  • noexit=vehicle To label a cul-de-sac or dead end street for vehicles (pedestrian can pass)
  • noexit=motor_vehicle To label a cul-de-sac or dead end street for motor_vehicles (bicycles can pass)
  • noexit=motorcar To label a cul-de-sac or dead end street for motorcars ((motor_)bicycles can pass)

-- 11:28, 27. Mai 2009 User:Vsandre

The problem of this new schema is that noexit has been tagged for a long time with the only description : noexit=yes To label a cul-de-sac or dead end street from wich it's hard to asume if people have thought of vehicle or any means of transport. I personnaly have used it by assuming a dead end for vehicles only (corresponding to the map features sign). This new schema suppose the noexit=yes means that not even pedestrians can continue. Sletuffe 15:41, 27 May 2009 (UTC)

Well that's what noexit=yes means. It means helicopters can't continue. --pmailkeey 2016:4:10

This is so stupid. We already have all this variants of which vehicles that may pass or not in the ways. How a about rollerblades? How about when driving a truck thats might not pass under a bridge? If users need to see this kind of dead end information on a map, it is obvious to me that it is a task for some map renderer to draw, for a given vehicle type, from the already existing data. We do not need noexit-tag. --Henriko 23:40, 7 March 2010 (UTC)


Osmarenders zoom level 17 just renders:

noexit=yes/true (red transparent cross bar at tagged node)
noexit=vehicle (orange ...)
noexit=motor_vehicle/motorcar (yellow ...)


noexit:foot=yes/true (red ...)
noexit:bicycle/noexit:vehicle=yes/true (orange ...)
noexit:motor_bicycle/noexit:motorcar/noexit:hgv/noexit:goods=yes/true (yellow ...)

--Mueck 12:35, 12 June 2009 (UTC)

Just added at z17:

noexit=no/false (green ...)
FIXME/fixme=continue/continuation (green ...)

--Mueck 09:24, 14 June 2009 (UTC)

when did noexit apply only to nodes ?

tagwatch europe for noexit shows a major usage of this tag on way and not nodes. When was it decided that it applies only to nodes now ? (given the front page of noexit=*Sletuffe 15:41, 27 May 2009 (UTC)

For this tag to be used on a way it has to be on a node of that way - so is always on a node ???? --pmailkeey 2016:4:10
This clearly wasn't a simple error. The "node only" information can also be found on Map Features and several German pages I know of, so changing the documentation on this single page doesn't improve anything. Instead, we should try to find out why and, more importantly, how people use it on ways. --Tordanik 15:01, 12 June 2009 (UTC)
My basic answer to this problem, is : if someone wants to change the properties/descriptions/apply to/... of an allready approved/existing/used tag, he has to go on the tag's talk page and explain his reasons before changing it unilateraly or else I don't see why I souldn't reverse his changes. If his point stands, and he proposes a smooth migration to new usage, then why not. But doing so will just make the noexit tag on the data become meaningless since it was tagged by users at a time where they considered it to be "apply to way". Sletuffe 16:20, 12 June 2009 (UTC)
I found the history of this change here [1] It seams there was an edit war envolving an old friend I knew quite well that used to change things without discussion. I'm returning to historical meaning, and be glad to start the dicussion with you. Sletuffe 17:03, 12 June 2009 (UTC)
So how can these ways tagged with noexit=yes be used right now? Even if we say that it can be used on ways, it's still under-documented because we don't know which way it is added to. The last one before the end node? All ways since the last junction? (Do only junctions with ways that are usable for vehicles count?) All ways since a traffic sign? Tagwatch doesn't help to answer this, and I suspect there are examples for every possible interpretation. ---Tordanik 16:55, 12 June 2009 (UTC)
The definition from December 2006, "A dead end road/cul de sac with only one access road", is unambiguous that it'd be used only on the way from the last intersection to the dead end. But it's ambiguous as to what transport modes that affects, (motor)vehicles or pedestrians too. Noexit=yes, when used on both ways and nodes, can be used to find the mistakenly connected nodes and deliberately not-connected-to-nearby-road ways, if the tag does mean that the only connected node is the only exit for all users. I personally haven't used it anywhere, except on nodes at ends of tram platforms, where there's no connection to the road really close to it. A linear barrier between them would mean the same, though, if there is one. Alv 12:33, 13 June 2009 (UTC)
I feel a bit just like it was discussed down there about the real goal of this tag. Either it is a internal tag (to say the mapping is not finished) or it might well be something that can be guessed automatically (does the street continue or not).
As of where to tag it I would say : once you enter a way tagged as noexit (by coming from another way) you know you'll have to turn back at some point.
The other solution would be to tag the sign and set it as a node, with some complex relation/restriction to say to wish street it apply.
(Strangely, my personnal choice would be just to deprecate this tag)

Sletuffe 17:15, 12 June 2009 (UTC)

It is actually ambiguous. what do do if before the dead end from the intersection there are 5 ways (because of different speed limits, surfaces etc) ? personally, i'm tagging this on the dead end nodes. --Richlv 16:00, 6 December 2009 (UTC)
That's a good point. I think we could change the page to allow noexit=yes to be used on nodes (at then end node of the way) and/or on the last part of a way that ends in a cul-de-sac. If noone feels hurt, we can add onNode=yes and explain that matter in the text. sletuffe 22:37, 6 December 2009 (UTC)
Hey this tag to nodes is really usefull, it indicates -- I got no further because there is no possibility, or if noexit=no it indicates, please continue mapping here. So noxexit=yes should be used for dead end streets, as long as there is no possibility for pedestrian/cyclists to continue. The usage as indicator for cul-de-sac if pedestrians can continue is in my eyes wrong however.
I don't need this tag inside of cities, but outside in the forest. There are often ways on yahoo imagery, that 10 years ago were clearly usable. Now the forest has taken the way back, so noexit=yes is clearly a needed tag. Also tagging nodes with noexit=no is needed, this enables mappers in the nature on a quick view where to go to continue mapping!!!!!!!! You can't deprecate a widely used tag. Also I am against tagging a ways instead of nodes with noexit=yes, because it is more difficult to render. What we want to render is a sign at the END of the way, not at the beginning. Tagging the whole way as noexit=no is 1. misleading, because how shall I know where the noexit=no is supposed to start, and second impossible for the renderer to decide where to put a point to indicate it, if another mapper connects the road whithout looking at the attributes (and yes, I would say this is the majority of cases).
If you want to have a cul-de-sac description, yes I agree, apply it to the whole road, and tag the entry point from which it has to be clear (or some meters inwards so that it is clear to which way it applies. However what the tag description says is that it indicates the state of mapping and whether or not I have to continue mapping the street, and yes this is a needed feature. When I go out tagging in the nature and come to a crossing, I go each way 20m inwards with my GPS, then continue my normal journey. At home I map the beginning of the ways as noexit=no at the LAST node, doing so every mapper can with a quick glance see that there is still something to map!--Extremecarver 19:44, 12 June 2009 (UTC)
Forest has grown enough to make it unusable: barrier=trees foot=yes or switch from track to footway.
Sorry but barrier=trees is stupid. Trees are no barrier per se. This is about ways that once existed that also on foot you can't proceed anymore any better than directly walking through the forest instead of using a way !
Okay, then there's no highway at all. The road/track/footway leading there just ends there. Alv 12:33, 13 June 2009 (UTC)
Road continues but I didn't map it: FIXME=continue
Why use FIXME|continue (which is not defined) if we have noexit=no which was already approved. We don't need creative new tags--Extremecarver 09:55, 13 June 2009 (UTC)
It's already more popular: in Europe fixme=continue (+ continuation) 1235, noexit=no/false 508. I found only two of noexit=no/false in other continents - both do convey the same meaning but the others will stay on the map once it's finished and the fixmes will be removed eventually. Alv 12:33, 13 June 2009 (UTC)
"I was in a car, but only on foot I could have proceeded further": a stub way with highway=footway and a FIXME=continue at the last node.
A footway between carriageways ends abruptly but there's no connection to get to cross the road: noexit=yes on the last node. The only case (well, similar might exist with other highway types) a validating tool can't distinguish from a missing continuation across the road. Alv 23:04, 12 June 2009 (UTC)
that would be correct.--Extremecarver 09:55, 13 June 2009 (UTC)

Is there really a discrepancy between "noexit at node" and "noexit at way"?

  • "noexit at node" may mark the physical end of road (for mapping as hint for users (distincting motor_veh./foot/...) and mappers (please continue mapping / is not a bug, it's a feature / ...)
  • ... or "place of traffic sign" at the beginning (to map?)
  • "noexit at way" may be a hint for routing: "The destination you want to reach is in a cul-de-sac, if you drive a 40 metres long hgv you should look for possibilities to turn before enter it ..."

The first one is (for my opinion) the most valuable form, that I may force with rendering at z17 in Osmarender ;-) but this doesn't mean, that the other ones are wrong, especially the last one may be valuable, too. The second one ... May be, it should be tagged as traffic_sign=noexit to distinct from first one? --Mueck 09:38, 14 June 2009 (UTC)

The signs definitively as traffic_sign=noexit or similar, if someone wants to. Alv 10:35, 14 June 2009 (UTC)
If used as inter-mapper information, I prefer the fixme=continue solution. fixme is made for that, noexit was probably used to indicate a cul-de-sac, juste like it was written in the access page. Sletuffe 12:31, 14 June 2009 (UTC)

Why do we need to tag a cul-de-sac

Is there any need to tag a cul-de-sac? If so, how should it be done? Smsm1 15:10, 10 February 2008 (UTC)

About cul-de-sac and noexit tag, are these vehicle dependant ??A street could be a dead end for a car, but have an exit for bicycle or pedestrians ... Is it a a valid noexit then ??? --PhilippeP 08:36, 8 April 2008 (BST)

At least in terms of German road signs you will find a dead-end sign at the entry to such a street so I guess a street qualifies as dead-end if there is no exit to cars. Ukuester 07:51, 3 July 2008 (UTC)
This tag should NOT be used for dead ends, but rather where a legal restriction prevents you from passing through, in Norway these places usually have a sign saying "Ingen gjennomkjøring" meaning "No through traffic". You can pass through the road ONLY if you have a purpose somewhere along the road, such as delivering goods to a property along the road, and than leave on the opposite side. But I don't think this really have to do with access, more to do with restrictions. I would use a restriction relation on it instead. --Skippern 11:00, 15 December 2008 (UTC)
For such restrictions, I use the tag access=destination -- Pieren 11:57, 15 December 2008 (UTC)
Indeed, that's where access=destination is for: only allowed to enter if you have something to do in that area. I can't see what it has to do with turn restrictions. --Eimai 14:07, 15 December 2008 (UTC)
Back to the begining's question (1 and a half year later) : Some people seams to think that this tag is useless and confusing. Probably because the existence of a dead end can be determined by the topology of ways without the need of an extra tag. Since it is still in use, removing this page doesn't seams right to me, but perhaps indicating it shouldn't be used anymore. (I personaly don't care, only reacting to recent deletion and moves) Sletuffe 16:03, 27 May 2009 (UTC)
I do think we need the tag. If a roads ends somewhere, how should I know from looking at the map if it simply ends there, or if this is as far as the last mapper got. If I see noexit=something I do know that I can't advance. I think however the above proposal of changing values is good. Maybe a way is dead end for bicycles, but not for pedestrians. Then adding noexit=bicycle would be helpful. Of course this can be encoded in the access tags too, but for map renderers it is much easier using the access tags to display a symbol.--Extremecarver 09:11, 3 June 2009 (UTC)
A way that ends on a map which marks a continuing road can be marked with 'fixme' - I've found these. I find roads that are mapped as ended yet continue in places where a validator wouldn't flag up the error. It has to be accepted that the map contains errors - and if you are familiar with noexit=yes then you are an editor and should fix the map correctly if possible - if not, tag with fixme. --pmailkeey 2016:4:10

Moved from Talk:Map_Features#noexit: 01:58, 15 June 2009 (UTC)

Whats the purpose of noexit?

In real life I can definitely see a purpose of signs telling people that a street is a dead end.

But I can not see any use of this tag in a map.

Both humans and routing algorithms will see this very easy anyway. --Henriko

I've pondered the same, but only lately I discovered one use for it in quality control: consider a dual carriageway urban road with tram lines and tram stops in between the carriageways, with both the footways and pedestrian crossings leading there and also those serving as tram platforms individually mapped. If such a platform way is connected to the rest of the map only in the other end, no data consistency check algorithm can know if the free-floating/noexit end is a mistakenly disconnected node (distance to nearest other highway way is 4 to 6 meters) or a dead end. Alv 21:37, 3 May 2009 (UTC)
This is very useful for mapping and for using the map. The OSM map is still incomplete in most parts of the world and in most areas outside of major cities. The noexit allows you to distinguish between a real dead end and a way that just hasn't been mapped completely yet. Your statement "noexit is very easy to see" assumes that the map is complete which is not true in most cases. --Nop 21:59, 3 May 2009 (UTC)
So you say that noexit should be used EVERYWHERE where a road ends? And then this should be maintained when places grows? --Henriko 00:07, 4 May 2009 (UTC)
What you say about the map is not finished, and that there are a lot of dead end look alikes is true. And it is something I have already thought about. I can not see how a routing algorirm or a human can trust a unfinished/unmapped road anyway. Only by knowing it is not a dead end, it is still quite stupid to head out that road, since you in most cases have no idea where it might take you. What about an unfinished/unmapped dead end road? Even if you actually could see a noexit notice on the map, because it will end eventually, you wont know if it is halfway to your goal, or if you can still reach your goal that way. And what about passing thru an area with a lot of mapped roads that actually have two exits but only one is mapped. Do you really think it is that helpful if the map could tell you that it is possible to drive thru a such area, but not how, so you might end up an hour or two searching for the unmapped exit? Or where should this be placed? As the street signs are? Or where the road ends? Im not getting it. --Henriko 00:07, 4 May 2009 (UTC)
One more thing that is a problem is the fact that what is a dead end for a car, or perhaps for a high truck under a bridge, is not necessary a dead end for a bike. This is, as I see it, the final proof that noexit is stupid, inadequate and redundant. --Henriko 00:43, 4 May 2009 (UTC)
I don't think I would use noexit for that. It could be useful in directions if only used where it is clearly the end of a road (which a dead end for a truck wouldn't be), but "where the road becomes a cycleway or footway, or has bollards across it, would be clearer. Peter James 21:34, 6 June 2009 (UTC)
Ok. I can see the use of something to guide map error checkers, in some cases. But then it should not have a roadsign associated with it, and it should not be rendered on regular maps. If we go in this direction, then it will be more related to tags like noname, and it should be moved out from the access group. Perhaps noexit is not a good name for it then. --Henriko 00:32, 4 May 2009 (UTC)
I agree it needs a better name - maybe end=yes or deadend=yes. It should probably be in the properties section. Peter James 21:22, 6 June 2009 (UTC)
Or possibly separate tags such as highway=end (and possibly road=end), railway=end and canal=end. Peter James 21:34, 6 June 2009 (UTC)
I do not see any good use for this tag. I will descuss three possible uses: 1. To render it on the map: There are many problems. Witch direction is the sign? (As some here points out) it might not apply to anyone else than trucks or cars. Often there is variants of this sign to indicate this. How would that be tagged? And most important: Has anyone of you seen a "No exit" symbol on a printed map, ever? I sure haven't. The actual sign is out there along the road for folks that don't have a map in the first place. 2: For routing: Useless (already discussed). 3: For mappers. We're much better off using the fixme tag for any information about what you know is wrong or missing on the map (incomplete roads etc). My conclusion: This tag is useless and should be depreciated Gorm 01:21, 15 June 2009 (UTC)
As allready mentionned, algorithm can guest which street is or isn't a dead end. The only use I see for this tag is as a mapper internal to say "not finished, please help in mapping further". Agreeing on a fixme=?? value could possibly do the job. (fixme=continue looks like a good candidate as it is allready used) Sletuffe 10:57, 15 June 2009 (UTC)
As already mentioned: This tag is useful when applied to nodes. 1. No problem, just render it exactly where the node is. As this usually is the end node of a way, there is no problem with direction. If the way continues as a different type, it obviously has an exit and should not be tagged this way. And yes, this tag is rendered on the hiking map and it is very useful to determine that that shortcut up the hill is actually a cul-de-sac and not just an unexplored through track. --Nop 13:36, 15 June 2009 (UTC)

recent edits (2009-11)

There have been repeated attempts to edit the page. Pieren's version

  • removes any mention of the tag noexit=no, which is still being used
  • can be read to alter the description of noexit=yes (from "The way ends in the forest or before a barrier. [...]" to "Don't use this tag for regular streets ending nowhere.")

One edit referred to discussion, probably #Candidate_for_the_most_stupid_tag_competition?. However, this discussion can at most be a reason to add a recommendation not to use this tag. It cannot be a reason to remove documentation for the tag while it's still being used. Therefore, my suggestion would be:

  • Leave the noexit=no section as it was before. It already recommends FIXME=continue.
  • Merge the introduction and the noexit=yes section to
"noexit=yes is used to explicitly indicate the end of a way.
Adding this tag can be useful when other mappers or quality check programs might assume an error otherwise, e.g. when a way is ending near other ways but is in fact not connected."

This solution wouldn't affect the meaning of the tag (after all, adding it to a way that doesn't end near other ways isn't wrong, it's just unnecessary), but clearly state its purpose. --Tordanik 19:35, 21 November 2009 (UTC)

Your suggestion sounds reasonable. You could even drop the noexit=no section in my eyes. But simply changing/removing the definition of a noexit=yes which has 56000 uses and people who are evaluating it, is unacceptable. --Nop 13:34, 22 November 2009 (UTC)

Noexit should not be used.

I've heard of people complaining about others 'mapping for the renderer' and here we're encouraging people to map for the validator - INCORRECTLY !

If it's not possible to get from one highway to another, rather than saying it just can't be done by using noexit, put the detail in that prevents it from being done. Is it just a kerb or is there a fence, hedge, wall or what ? It is that that should be added in, not just simply say something can't be done. --pmailkeey 2016:04:10

Useful for railways and waterways?

The above comments seem to indicate we're headed towards consensus that noexit=yes is useful as a tool to silence false positives in validators. Why would this be limited to highways, though?

Railways may end without a buffer stop or similar, and while I don't know if validators check that, the logic would be the same as for highways.

Waterways are supposed to end at another waterway, and validators do check that. However, there are cases where it's impossible for mappers to know where some waterways continue, e.g. entering the city sewer system, so there should be a way to silence warnings when it's a false positive. noexit=yes already exists and seems like a reasonable solution. StephenTX (talk) 20:55, 21 November 2016 (UTC)

Anyone using this tag, in 2017 (or later)?

I understand that it has been put into the map data, a "few times". noexit at Taginfo. I also understand that JOSM have some support for it. But I don't see any support in tools like OSM inspecor and Keep right.

Has anybody really used this tag, ever, except when testing stuff during arguments about its existence?

Is there any real practical uses for it, so I should bother to put it in on every regular dead end street? Or can it just be ignored until one run into some rare case of verification error? Does anybody care at all 2017?

Just wondering, since I haven't seen any autogenerated warnings, or an angry messages from humans, in about 8 years without using it. [unsigned, by Henriko according to diffs]

I don't know how to find usage from taginfo over time, but I think a tag with ~460k uses by ~15k unique users counts as "in use". I use it only to silence JOSM validator warnings about a way that ends near another way when they're false positives, which comes up fairly frequently; true positives (due to human error or bad imports) are common enough that I can't just disable that warning entirely. IOW, I treat this just like the "noname" tag.
I personally disagree with the other usage and therefore any rendering it may seem to imply; routers can figure out if a way is useless easily enough, so explicitly tagging it is just clutter--and adds a potential source of error if someone gets it wrong. Of course, so few (as a percent) ways/nodes have it--on top of the disagreement about what it means--that it's unlikely routers or renderers will ever bother, which makes it even more useless. StephenTX (talk) 22:34, 8 August 2017 (UTC)
Yes. Mapping it, rendering it on the map, using it for planning and considering it one of the advantages of OSM maps over ordinary maps. --Nop (talk) 18:27, 9 August 2017 (UTC)
If the map is "finished" in the area, how can it possible be more self explanatory if the road just ends, rather than putting some symbol at the end?
If the map is not "finished" in the area, how can it be useful for you to know that a road perhaps continues a few 100 meter more, or perhaps continues kind of forever, but you don't know where, and you don't know what kind of road it is? How do you plan a trip along such a way not shown on the map?
If you have a way for buses only, a bicycle path where pedestrians are allowed, and a staircase, all ending in the same node, how would you tag that node, and what would your dream renderer make out of it? --Henriko (talk) 22:06, 9 August 2017 (UTC)