No idea. That definition only describes half of what makes a police station, anyhow. When this location opened, it was described variously as an office or a station. Unusually, it belongs to a private nonprofit organization. I’ve only been by outside business hours and thankfully have never had business to report there.
There’s no need for vacant:shop=yes as “generally clean and in good condition”. disused:shop=yes can be used for that, which you already said “minor disrepair” is only a “may be”. Precise quality have eg operational_status= / building:condition= (when using them properly somehow) as an attribute, if needed.
What vacant:shop= can be necessary for is complete vacancy, an empty store space; whether newly built, or totally cleared out. This isn’t described by anything other than was:shop= + proposed:shop= , which could yet mean there’s technically not even a store space now.
On the other hand, this could still be disused:shop=yes
That’s a lot of replies and information.
So far I am receptive to vacant:shop=yes, but seeing these new information and to refocus discussion (section break / summary), I am gonna restate the original question:
What should we use to indicate the “zero”/“null” value of a shop?
We see lifecycle proponents of disused:shop=* but this being inside disused:* implies something was there previously. This does not hold for new shop spaces that simply hasn’t been used yet, or just an old empty shop space whose history could not be confirmed/established.
We also see shop=vacant . This feels natural when it comes to programming and data structure (this is a shop-type but is currently not initialized), but then this is classified as a trolltag (understandable), since it negates the meaning of shop=*. The general public can’t actually buy anything there because the shop/store just doesn’t exist.
I feel like the ideal schema should be like this:
- “This shop is null” (we are here; what tag-value to use?)
- If the shop is known to have previous usage, then add e.g.
disused:shop=*orwas:shop=* - If the shop is known to be a faker (e.g. police honeypot / undercover), then add e.g.
mimics:shop=*
Some extra perspective: land that is currently known to be unused (i.e. “null”) gets either landuse=greenfield or landuse=brownfield, depending on the exact situation. Could we have something similar to shops / shop spaces?
This increasingly feels like something that could be escalated to be discussed with the existing lifecycle tagging, but we will see.
2 Likes
That’s a good point. The more you look around, the more you realize that these placeholder values were ubiquitous before we developed namespaces or this notion of “troll tagging”. The only practical issue I see is just that we’ve gotten so used to treating shop=*, office=*, and craft=* as purely classification keys rather than thematic keys. This was strongly influenced by renderers, though some like OSM Carto are starting to take a more discerning approach. Like landuse=brownfield and landuse=greenfield, shop=vacant has been so common for so long that it’s hardly a surprise, except to someone developing brand-new software from scratch. In that case, there are so many other surprises in our tagging system…
Whenever data consumers use wildcards instead of looking for specific keys or values, mappers don’t have to go through so much bureaucracy in order to establish new values for less common cases. This has been fantastically beneficial for the development of shop=*, office=*, and craft=*, as well as for namespaces like name:*=*, addr:*=*, and payment:*=*. On the other hand, it narrows the keys’ semantic scope, so that we all have to read the name of the key very literally. Some edge cases are bound to fall out of scope.
A disused storefront that was most recently an OSM amenity can still be a shop=vacant.
That was part of my point in my post about all the things that go into the shop key in OSM but aren’t traditional “walk in and buy thing” stores…
A retail space can readily be converted from a coffee shop (amenity in OSM) to a convenience store (shop) and then back, or from an insurance broker’s office (office) to a prep school or a driving school (both amenity). A slight format change can change a space from an amenity=cafe to a shop=pastry… etc. And in between all those they can be vacant storefronts.
I probably wouldn’t tag a purpose-built former doctor’s clinic building as a shop=vacant, but there’s definitely OSM amenitys that become shop=vacant when closed.
E.g. Node History: Dear Grain (4349217185) | OpenStreetMap went from being a cafe (before it was in OSM) to being a chain coffee shop to being a bakery that also serves pastries and coffee, but because the bread is primary it became a shop=bakery. If it changes to another business that doesn’t put the expensive sourdough breads front-and-centre it might become an amenity=cafe again.
Node History: Batibot Filipino Foods and Grocery (6493338036) | OpenStreetMap went from being a payday loan chain (shop=money_lender in OSM) to being a hot food takeout (amenity=fast_food).
Node History: Shangri-La Fruit Market (530804330) | OpenStreetMap was a greengrocer, had a brief stint as a fast food kind of place (making a show of making food on premises), then went back to a greengrocer.
Node History: Osprey (6493214401) | OpenStreetMap went from being a shop=clothes to a shop=tattoo to a coffee shop (amenity=cafe). If the coffee shop closes, it’ll become a shop=vacant again.
4 Likes
Moreover as can become non-residential uses like dentist, gym. or micro health place in most places easily! Prefer building =yes pray that it’s mixed use vertically and then if retail vacant delete the point for that at will!!
I think it’s also OK to just not map things, for example in the case of some new vacant place. If we were to map it, we’d probably assume (maybe erroneously) that a shop will be there (as opposed to, e.g. an office) and we’d be kind of “mapping for the mapper” (i.e. a placeholder only useful for contributors to eventually map something more interesting. But we can always argue about usefulness of data to consumers! “What about my website for small businnesses who are searching for a place to rent?” 
).
If the idea is to precisely map an indoor layout, one could use room=*.
I think the flip side is that, if we don’t have an obvious way to map a vacant space, mappers will gravitate toward other means of indicating the same information that we might not like so much, such as naming the address point “Former Starbucks (closed)” or the building “(Vacant)”.
Not everything people want to say about their surroundings is going to be useful to a competitor to Google Maps for Android, but it could still be useful for other purposes, such as comprehending the increase in vacant retail space in a declining business district. If nothing else, the address (including unit number) continues to be valid present-day information.
I wouldn’t normally set a vacant space’s website=* to a rental listing website. However, many property managers maintain websites for their office buildings. The websites typically list available units, along with the building’s amenities and history. I don’t have a problem with linking to these sites, as the buildings’ official sites.
3 Likes