From OpenStreetMap Wiki
Jump to navigation Jump to search

what's wrong?

What's wrong with shop=vacant? A vacant shop looks like a vacant shop, and we tag what we see. Adding disused=yes or similar won't do the trick, because a shop loses its classification (clothes, food, toys...) as soon as it becomes vacant. It still does look like a shop, but usually it cannot be seen which kind of shop it was. --Fkv (talk) 22:47, 19 September 2014 (UTC)

Well, let's repeat what I wrote on Template talk:Map Features:shop: For me, a major argument against using shop=vacant is that it contradicts the definition of the shop key ("A place selling retail products or services."), as vacant shops don't sell anything. From a more practical perspective, it may be a good idea for certain applications to show all shops. With thousands of different shops, you really should be able to just filter using the shop=* key without having to deal with not-actually-shops.
And if the classification bothers you, there is still the option of using disused:shop=yes. --Tordanik 19:46, 21 September 2014 (UTC)
I used shop=vacant a lot, and never really thought about it. I can see disused:shop=* has some advantages although I don't feel strongly either way.
A vacant shop still has some of the interesting characteristics of a shop when it's taking up a space on the high street (it's that kind of thing I'm imagining). In a way you could say a vacant shop is still a shop. Note however, I would not say a vacant butchers shop is still a butchers shop. A data consumer is probably not going direct somebody to it in the hope of buying something because we're saying nothing about the shop type. It's not too bad on the "principle of least surprise".
On the other hand, yeah maybe 'disused' is better.
Most of the times when I've mapped vacant shops, it hasn't been particularly apparent what type of shop it was or will become. So I guess disused:shop=yes captures that. Shall we document that on this page?
-- Harry Wood (talk) 14:30, 24 February 2016 (UTC)
I also used shop=vacant a lot to map vacant shops, and I feel like it is the right thing to do. To me, a vacant shop and a disused shop are two different things. The latter makes me think of an abandonned, discontinued commercial space that has been empty for a longer time and is not readily available without major construction work for a new business to move in, a place that the owner has obviously given up, falling into a ruin or being demolished soon. As a vacant shop I picture a commercial space that is generally in good, maintained shape and ready for a new business to move in, often signposted "for lease". Aside from that, I think the definition of a shop "A place selling retail products or services." is much too narrow. The word shop may refer to a business that sells things but it can also refer to some commercial real estate, like a building or a space in a building. The latter idea of shop would also agree to the motto "we map what we see". We see the tangible shop (empty or not) rather than the intangible business. The unspecific "shop=yes" is pretty useless. I might care for specific shops, even for vacant shops if I'd be looking to open a business, but I cannot imagine why I would be interessted in a shop without knowing what it sells. For that reason I think "shop=vacant" is as good of a shop value as any other - except of shop=yes. -- Freetz 08:03, 19. September 2017 (UTC)
In OSM tagging, there's a distinction between "disused" and "abandoned". The situation you describe (an abandoned commercial space that is falling into disrepair) seems like a perfect job for abandoned, while disused is available for a vacant but still usable space.
To me, shop=vacant is pretty much a trolltag (or "troll value", as the case may be). "This is a place selling retail products or services. More specifically, it sells ... nothing." You may prefer a different definition for the key, but unless there's a consensus for that change, the current definition is still what data consumers should be able to rely on. We don't need to speculate about possible use cases, either, as displaying a generic icon for unknown shop types (e.g. a dot) is something that is already being done in practice. --Tordanik 09:38, 19 September 2017 (UTC)
Why would data consumers not be able to handle shop=vacant in a specific way? This value is documented in the wiki and used over 13,000 times. Any consumer that is serious about shops, should already handle this in whatsoever way the developers think is best for their application. Sure, there is a more or less subtle difference between "disused" and "abandoned", both in OSM and IRL. However, the real-life "disused" refers to a more final decomminsioning, usually leading up to either "abandonned" or "demolished". A vacant shop is neither "disused" nor "abandonned", it is more like "temporarily out of service" and expected to be back in service anytime. This aspect is what brings "shop=vacant" even close to "shop=yes". As we all know, OSM is not real-time media. As such, a vacant - or in your preference disused - shop might have already been re-opened by the time a users looks at OSM or at any consumers - just the same way any "shop=*" might have ran out of business - unnoticed in OSM - by the same time. Most consumers, however, ignore "disused" items. Hence, a disused shop is very likely to be hidden from most users - even though it might be a "shop=yes" in the meantime. If you did consider "shop=vacant" as a shop that may or may not sell (specific) things, it is as useful as "shop=yes". In my view, "shop=vacant" comes with the advantage that it is found by most consumers contrary to disused. OSM neither promises that a shop=chocolate sells chocolate at any specific time, neither does it promise that a shop=vacant sells nothing at any specific time -- Freetz 09:04, 21. September 2017 (UTC)
One can also argue if using the defintion of "A place selling retail products or services." and "disused" makes sense. A place doesn't sell things, people do or machines do, in general a business. A business, however - as a legal entity - cannot become disused. After it ran out of business, it is gone. -- Freetz 09:32 21. September 2017 (UTC)
I agree with Freetz (and Fkv); see my comments below under landuse. Johnparis (talk) 09:54, 19 September 2017 (UTC)
I hear the argument that shop=vacant could be a trollotag since shop=* describes a shop and an empty commercial space is not a shop=*.
so what is the right tag to describe an empty commercial location e.g. several rooms on the ground floor of a building with a sign "commercial space for rent" ? imho the right value would be shop=no, and this should not bother the users of the data since it is a basic rule--Marc marc (talk) 13:55, 17 February 2021 (UTC)
For me it is a good usecase for lifecycle prefixes - shop that ceased operation, with remaining traces is a completely different category (unlike quarry landform that stopped operation or unused building). I would use disused:shop=supermarket or disused:shop=yes 17:11, 17 February 2021 (UTC)
shop=vaccant for me doesn't describe the past nor the future, it describes the current state : it's an available commercial space without an active shop. if there was no shop before, disused:shop=yes doesn't fit, so the solution I see is a vaccant:shop=yes : today it's vaccant/available for a store instead of referring to the past. but to create a namespace almost only for shop and office, it's unlikely to be used Marc marc (talk) 21:28, 19 March 2021 (UTC)
If shop is not yet open it is also disused (for now), right? Mateusz Konieczny (talk) 10:30, 20 March 2021 (UTC)
No, "disused" means no longer in use, emphasizing that something was used before. There's unused, in-use, and disused - in this order. That's why I don't favor to use "disused" for something "temporarily not in use" or "vacant". Freetz
@Mateusz Konieczny, I came across your recent change. To rule out misconceptions of the meaning of the word "disused", let's clarify: "Disused" is very different from "unused". While "unused" may refer to something that has never been used and is "new", it may also be used for something that is "currently not in use". "Disused" is different from both meanings of "unused" and always expresses that something was in use before it became "unused". "Disused" also implies a somewhat longer expected period of non-use, most of the time probably a final decommissioning. A disused item, in most cases, will also be something that shows very significant signs of usage, wear or aging. The difference between "new and never been in operation yet" and "disused (not new and been in operations before)", may seem minor to some. Let's think of a new toothbrush, of an unused toothbrush, and of a disused toothbrush. They are all different. To me, that difference is relevant if someome offers me a toothbrush for sale. I might buy an unused one, after confirmation that it's new, but I would decline the offer to buy a disused toothbrush. The wiki shouldn't be defining relevances of differences and deteriorating the language. Let the user decide what to make of the data. Regardless of the syntax, a disused:shop=yes, or respectively a shop=disused, is semantically not the same as a shop=vacant. I find it unfortunate that many mappers use them synonymously; it doesn't make the data better quality, IMO.
Btw, over at Wikipedia some individuals seem to take ownership of articles; they constantly monitor and then improve and disapprove edits by others, let's leave those habbits with that project and let's collaboratively improve articles here. -- Freetz, 11:53, 9 September 2022 (UTC)
shop=no doesn't make sense to me (despite it has been used in a different context in the past). This could be applied (and probably would be applied by some mappers) to anything that is not a shop, being it a residential building, a restaurant, camp grounds or even traffic lights. Syntactically, there is no difference between shop=no and shop=vacant, the latter contains more information though and seems more logical to me. I don't see any good reason to use shop=no over shop=vacant. I find shop=vacant is the most intuitive and easiest thing to use and the most preferable for this reason. Making a problem out of the fact that a vacant shop doesn't sell anything is a troll argument to me. What specific kind of a shop a vacant shop is and what it sells is widely understood and this is perfectly defined behavior. After years, I still don't see any (practical) problem with shop=vacant for an end user or data consumer with this. I even think it has advantages over the life cycle tags because its ease of use. Semantically, there is no difference between disused:shop=yes and shop=disused. The difference is in the syntax. And a good syntax is simple. Replace "disused" by "vacant" and we arrive at the de-facto standard shop=vacant :) -- Freetz
Resolved: article seems to be describing this now Mateusz Konieczny (talk) 11:27, 4 March 2022 (UTC)


I'm sure I had a better photo of this. I'll have to dig it out. Trouble with the current one is, it's not actually entirely apparent that the shop is vacant. It might just have it's metal shutters closed (which is a common problem while mapping shops actually) -- Harry Wood (talk) 14:33, 24 February 2016 (UTC)

I updated the photo. Johnparis (talk) 09:55, 19 September 2017 (UTC)
Resolved: Mateusz Konieczny (talk) 10:43, 4 March 2022 (UTC)

shop=vacant is really redundant

Many small vacant shops are still used as apartments in the floors above so it is not really vacant or disused. Why not delete the shop relevant tags and just leave the building attributes as is?

-- User:Dcp 14:06, 21 August 2016

In case where there is no vacant shop (because it is now used for something else), this tag should be not used Mateusz Konieczny (talk) 11:26, 4 March 2022 (UTC)

Land-use planning

In some cases, local land-use planning rules (such as a Dutch "bestemmingsplan") may require that the next user is also a shop. In that case, shop=vacant (or disused:shop=yes) may be more appropriate or informative than just building=yes, especially if the building is clearly recognisable as a former shop. --IByte (talk) 20:34, 9 September 2017 (UTC)

Using shop=vacant for this still has the problem that applications will often (rightfully imo) interpret any shop value as an active shop, not just a building mandated to be used for a shop by land-use planning rules. I believe disused:shop=yes would be a great fit, though! --Tordanik 16:21, 10 September 2017 (UTC)
I don't see any end-user problem with shop=vacant. What do you envision, a journalist who is searching for "a shop of any sort", to interview a shopkeeper, and lo and behold when he arrives at a "vacant" shop he is disappointed? In the real world, people are looking for particular kinds of shops. And searching for vacant shops can have many useful purposes.
There is no tagger problem either; we "tag what we see." We don't tag for applications. I agree that the definition of shop is rather cramped, and could use revision, but that seems unlikely.
A greater problem for land-use is that it doesn't apply to nodes; in this case a generic landuse=commercial or even landuse=retail might be more descriptive, but that's not an option.
I thought of building=yes combined with disused:shop=yes, but that's not really accurate, as often there will be several storefronts in one building, only one of which is vacant. So what top-level tag would you use instead? shop=no? Kind of ridiculous.
So all in all, I'd say shop=vacant is the most useful top-level tag, possibly in combination with disused:shop=yes. Johnparis (talk) 09:52, 19 September 2017 (UTC)
I don't know why you believe that disused:shop=yes cannot be used on nodes on its own, but you're mistaken. It is a top-level tag, and can be used in any place where you want might consider using shop=vacant.
Furthermore, there is no such rule as "we don't tag for applications". In fact, making OSM data easy to work with for developers is a significant factor when designing tagging standards! You may be confusing it with Tagging for the renderer, which speaks out against using incorrect tags – precisely because doing so would break the documented definitions of tags that data consumers need to rely on.
And as for possible use cases, even the standard style has a catch-all dot for shop types that it does not have specific icons for. The user of the map is then shown a generic icon along with a name or other relevant information – surely that's still useful? --Tordanik 11:49, 19 September 2017 (UTC)
"I don't see any end-user problem with shop=vacant." for example OSM-Carto needed to add special rule in rendering of rare shop types to exclude it. Not a major issue, but it is one of many slightly broken parts of OSM tags. And note that shop:disused=yes can be tagged on a separate node, it is a top level tag Mateusz Konieczny (talk) 13:22, 27 November 2019 (UTC)
This is a problem of the data consumer and interpreter in my view. Carto needs (and has) special rules for many shop types anyway. In the data consumer perspective, a vacant shop is (or, should be) just one of many specific shops. Effectively removing a data item by changing its key name because some consumers do not have a specific rule, is the programmatic version of "tagging for the renderer". shop=vacant is not a broken tag but is valid data. What to make of it is the responsibility of the consumer. -- Freetz
Thanks for the correction. You're right, I was thinking of tagging for the renderer, which doesn't apply here. While Overpass Turbo and JOSM both allow searching for disused as a tag, neither ID nor Potlatch show it as a top-level tag. I imagine those are the most commonly used editors by taggers. I'm still not persuaded by your end-user case, since the "other relevant information" displayed would presumably, in this case, be "vacant", which is unambiguous. In any case, the good news is that people have the choice of using either or both. Johnparis (talk) 00:06, 21 September 2017 (UTC)

Not a placeholder

Although shop=vacant has value as a placeholder, there are numerous other (significant) use cases:

  • Counting the proportion of currently vacant shops in a given retail area. This metric is very widely used by local authorities, social scientists. It was one of the first things I wanted to do after I'd done detailed shop surveying: to compare my figure for vacant shops with other published figures.
  • disused:shop is not absolutely cognate with shop=vacant. For instance the place may be disused as a shop but currently used for something else (e.g., politicians office, micropub etc). In many places shops have been converted to living spaces in a property which would have once been the shopkeepers residence as well as business (I guess the argument is that these should be abandoned:shop=*, but as ever the lifecycle prefixes fail to really provide the ability to discriminate all cases).
  • disused:shop may still be relevant when an adjacent shop takes over the space of a shop which went out of business (something which has just happened with one of my local shops). In such cases shop=vacant objects would cease to exist.

The placeholder aspect is certainly important for anyone surveying shops, so I would not downplay that aspect. It is particular important to assist when re-surveying shops to find changes. SK53 (talk) 14:41, 26 November 2019 (UTC)

"the place may be disused as a shop but currently used for something else (e.g., politicians office, micropub etc)", "adjacent shop takes over the space of a shop" - in such cases it no longer qualifies for disused:shop nor shop=vacant. Note that disused:shop is not for "there was a shop here once in the past", it is for "there is a shop infrastructure here (special furniture/signage/etc) but shop is not open". In such case proper solution is to delete object, in rare cases removed:shop=* may make sense - where there is a real risk of remapping such object. Mateusz Konieczny (talk) 13:19, 27 November 2019 (UTC)
You can use was:shop=*. --- Kovposch (talk) 10:14, 4 March 2022 (UTC)



What you think about inactive:shop=* tagging for shops which are for example waiting for (re)opening so strictly speaking disused:shop=* would not fit? Mateusz Konieczny (talk) 16:59, 17 September 2022 (UTC)

I tag shops that are no longer vacant but where it was posted or announced otherwise that some business will open in it as e.g. planned:amenity=cafe, construction:shop=backery, or even tagged it completely as shop=bicycle with all its available info such as name=Mike's Bikes etc, and opening_hours=Mo-Su off - depending on my mood and the progress of the preparation for the new use. I usually add a fixme=* asking for re-surveying and updating once it's open such that this POI gets more attention by mappers. If I know a planned opening date, I include that, too.
I don't know what an "inactive shop" is, and I question that many other people do. Wording that isn't intuitively and commonly understood in a somewhat unique ways but needs an explanation will never be used "as intended". For shops that are in between vacancy and re-opening, I prefer the above schemes as they describe accurately in a language that is commonly understood what the situation of the shop is. If you're aksing about inactive:shop=* as an alternative to shop=vacant, I don't think we need any. The semantics in shop=vacant is just right and accurate. "Vacant shop" is as well understood and is as well in common use outside of OSM as "vacant hotel room" and "vacant job position". So if anything, as said before, the syntax could be questioned. Now wether one would suggest a new namespace vacant:shop=yes just for shops - which wouldn't make much sense to me (unless someone wants to make OSM OpenBooking, OpenAirBnB, and OpenMonster) - or create amenity=vacant_retail_premises, both in my view would be beating a dead horse. The syntax of shop=vacant is well established and many years old. Every consumer who had initial problems with finding "shops where nothing is sold", have either adapted to reality by now, or (should have) disappeared.
The root cause of the confusion, IMO, is the understanding of shop=* in the Wiki "A place selling retail products or services.". As I wrote before, a shop in English also refers to the premises (or even "originally" instead of "also", to my knowledge). An empty, vacant, disused or even rotten shop is still a shop. A "place" doesn't sell anything, people or machines do. To me, a re-definition of shop saying something along the lines of "A place intended for offering retail products or services" would make most sense. If, and only if a shop is defined as a particular shop, such as e.g. shop=bicycle, this should mean that bikes are offered at the POI. This is always the case - even for a vacant shop since the the shop itself is what is being offered. The redefinition wouldn't solve any significant problem though, and I still don't see any practical problem neither with the current definition in the Wiki of shop nor with shop=vacant, and I haven't been shown differently.
To me, disused:=* and abandoned:=* are flawed semantically as I argued over at* (Not just "strictly speaking". This isn't splitting hairs; it rather appears that some english wording is being used in very peculiar ways that are probably not commonly understood. I just think OSM Wiki shouldn't use a language that looks like english but actually isn't and defines its own meanings instead.) Freetz (talk) 11:16, 19 September 2022 (UTC)