Hi DaveF and others, I’m the person referenced at the start of the thread. I think it’s useful to have a wider discussion on this topic. I’ve always understood OpenStreetMap to have an on the ground rule where you map what you can see on the ground. There are multiple issues that this thread raises.
There are a handful of examples that have come up about how to map the data: 1. Schools and nurseries - often have multiple FHRS inspections due to the main school, nursery, breakfast club, and other groups all have separate inspections and FHRS. Sometimes the catering is an external caterer, but there is no on ground sign of this. The cater won’t cover anything other than the school - it feels like an implementation detail about how school meals are delivered to the children. 2. Pubs with separate restaurants - sometimes it’s a clear separate caterer, sometimes the separate certification is the only way you can tell that they are separate is maybe a different payment terminal to the main bar or the open data having separate certificates. 3. Convenience stores that are not open to the general public, instead only available to staff in a particular place. The only way to verify their existence is through the FHRS open data and the store locator, however there isn’t an open license to this as far as I’m aware. As a member of the public we can’t see the existence of the store, unless we happen to work or otherwise get special access to the location. Surely Tesco should have a warning on that store page to highlight that the public can’t access it! 4. Places that don’t have a single main purpose or are shared use - how to adequately tag these - some examples: a) school with a nursery b) school with sports facilities that are open to the public - how to specify term time in opening hours without needing to manually update all the time? c) tell the difference between pub that do and don’t sell food (I see the food=yes tag plus the relevant opening hours for food/kitchen as the best option) d) a shop that sell computer games and outdoor clothing as that’s what interests the owner e) a cafe/beauty parlour that’s aimed at teenage girls f) a community centre or church hall which has a number of groups, but no external signage about those groups - each group has a separate FHRS id g) a high school and primary school are neighbouring on a site and hard to see where the boundary is h) school and community centre share a site i) multiple banks sharing a banking hub 5. The same FHRS id being mapped to multiple items. Looking at Taginfo this seems to have happened a number of times, some of which I suspect are errors, others are more likely where a business is spread out. https://taginfo.geofabrik.de/europe:united-kingdom/keys/fhrs:id#values Should entities that are available in some import but have no on the ground signage be mapped separately? Are they an implementation detail of another place? My view is that these shouldn’t be mapped separately as they are an implementation of something else. Could possibly add a tag on other things in this case eg the school? Or leave it as an external analysis with merging multiple data sources. Not all data needs to be stored in OSM. On the topic of wanting to track the number of companies that are externally employed to supply schools, could this be done via a tag on the school rather than needing a separate entity for it? Should it even be in the OSM DB, or should it be a separate analysis merging other data sources? The tag craft=caterer I would associate with the base of a company, such as a central kitchen and office for going preparing the food to go elsewhere for outside events at other venues. I would not associate it with a contractor at another place, so what other tag could we use for this? Could something like contractor=catering; catering=school_meals or similar maybe be an option. The other option would be on the school having things like catering_contractor:name=X catering_contractor:fhrs_id=Y which would be on of the semi colon separated values from the main list which is used for matching to the main DB, but allows other analysis to be done. Or could there be some form of external merging of the datasets to do the analysis. For a shopping centre where the public can access it, then yes all the individual shops should be mapped and FHRS ids attached appropriately. I’m of the view that we don’t need to have a one to one match between objects in OSM and other external datasets, there may be cases there there will be a one to many match. Another example would be where the same venue has got a new certificate and they haven’t taken down the old one. I’ve seen one case where there seemed to be duplicate FHRS ids with the certificate on the same day. I’ve heard of one mapper who managed to get the FHRS data fixed by emailing the local authority. Thanks, Shaun > On 20 Mar 2026, at 22:42, Jez Nicholson <[email protected]> wrote: > > It's a bit like pubs. Around my way, they have an FHRS:ID for the wet sales > (the bar) and a separate FHRS:ID for the food, which is often subbed out to > someone else to provide. I end up with a separate node for the food, which is > fine by me. > > On Fri, 20 Mar 2026, 22:32 Ken Kilfedder, <[email protected] > <mailto:[email protected]>> wrote: >> I think you're right. There shouldn't be multiple FHRS IDs for a single >> entity. If there are two businesses in the same building, they should have >> two differrent OSM objects anyway, and if they're both on FHRS, then the >> next steps are obvious. >> >> >> >> --- >> https://hdyc.neis-one.org/?spiregrain >> [email protected] <mailto:[email protected]> >> >> On Fri, 20 Mar 2026, at 20:03, Dave F via Talk-GB wrote: >>> Multiple FHRS:ID values on one entity. >>> >>> I've been in conversation with someone who believes adding multiple FHRS:ID >>> values to one entity is the correct way to tag. >>> >>> For example, a school where, not only does it have it's own FHRS:ID, but >>> the local authority employs external companies to provide facilities, such >>> as school meals. >>> >>> This creates the problem of being unable retrieve data on the number of >>> these companies from the OSM database as they're not explicitly named. >>> >>> He provided an example to justify his method, but I think it reinforces my >>> point - RBS HQ in Scotland: >>> >>> https://www.openstreetmap.org/way/91156737 >>> >>> This includes a Tesco express: >>> https://www.tesco.com/store-locator/edinburgh/175-glasgow-rd >>> https://ratings.food.gov.uk/business/661909 >>> >>> He claims it should be assigned to the building "as RBS building is a >>> private business that we can't easily map the inside of & the shop >>> inaccessible by the public or customers". >>> >>> Again, this prevents an accurate count of Tesco's shops. >>> >>> I perceive the employees as the customers, and giving the shop its own node >>> & tags inside the building is still more accurate & detailed, even if the >>> location within the building is rough guess. >>> >>> Am I missing some blatant logic? >>> >>> Cheers >>> DaveF >>> >>> _______________________________________________ >>> Talk-GB mailing list >>> [email protected] <mailto:[email protected]> >>> https://lists.openstreetmap.org/listinfo/talk-gb >>> >> >> _______________________________________________ >> Talk-GB mailing list >> [email protected] <mailto:[email protected]> >> https://lists.openstreetmap.org/listinfo/talk-gb > _______________________________________________ > Talk-GB mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/talk-gb
_______________________________________________ Talk-GB mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-gb
