Hi Martin I cooked up a suggestion for implementation of permids :)
Den Sat, 22 Aug 2020 20:56:05 +0200 skrev Re: [OSM-talk] New API suggestion: Allowing contributors to easily track their OSM-objects over time: > On 2020-08-22 19:55, Martin Koppenhoefer wrote: > > What kind of permanent ids do you want? For some more abstract > > concept like a road with a specific name? A shop? A building? If > > there’s a way tagged with building=supermarket, shop=supermarket, > > name=Foo, and you add a permanent id to it. Then the supermarket > > gets bought by someone else and changes to name=Pango? Or the > > supermarket closes. Or it gets torn down and rebuilt by the same > > operator. Or someone detaches the shop tags and adds them to a node > > inside. What happens with the id? > > Your comment is spot on. We're discussing technical ideas for > something that isn't even clear on a semantic level. > > Somehow this whole discussion reminds me of this blog post [1], in > particular the "Lack of Permanent IDs" section. A "Conceptual Object" > as it is described in that blog post sounds like a great idea, yet it > suffers from the same real world semantic issues that Martin already > pointed out. Could you give an example of the issues you are talking about? I'm thinking we have a need to preserve the ids of anything with tags in OSM (that is what you navigate via or to/from and whats interesting for the outside world to link to (I have seen very few links ever to nodes without tags, because they are often part of a way or relation with tags or useless orphan nodes)). This is how it could work: Say a shop: Mens Wear within a building is represented today as a node in osm with a permid=Z123. Then someone deletes it erroneously when the shop closes and we preserve Z123 and the last known position and node tags (because it has a permid). Someone else comes along and creates a new shop node: on the same spot or close nearby and now the question arises, should the new node created be linked to the old Z123 or should we create a new permid=124? We ask the mapper. If they get information about what was deleted, they can judge whether the Z123 is still conceptually valid and restore the old shopspace and rename it to the name of the new shop: (this would preserve the history old->new also from the earlier deleted node too essentially making it possible to restore nodes with permids because they are valuable). The permid then no longer represents a shop but the location of a space where a shop could and now does exist. Someone making a service cataloguing all shopspaces for hire in a city could then link to this shopspace FWIW. External tools like Wikidata might link to it because they want to list all locations for a particular brand of stores. If the shop moves to another shop area the permid of the shopspace remains the same, wikidata will then have to purge the location and could do that either by looking at the name tag present or if a qid is present. We could create a special API or egress dump with an entry every time a name or qid changes of one of our permids, that wikidata could ingest. This is all just spun up out of the top of my head so it might not be feasible, but I think it would work. Of course permids should be handled in the background when users are creating a node and adding a tag (I never get to select or edit a permid in wikidata but I can merge two who represent the same physical concept in the same spot and the earliest one (lowes id) lives on and the other higher id simply refers to it). I hope this made sense to you all, if not feel free to ask. > > > [1] https://blog.emacsen.net/blog/2018/02/16/osm-is-in-trouble/ Thanks for the link. I read it before and forgot about it. Reading it again I see 2 years have passed and not much has changed it seems (besides we have a little more AI stuff going on in HOT and FB and OSMF has opened for free membership of active mappers). Maybe he is right that OSM has lived out its time and is unable to adapt, maybe not. Lets create a high quality database, not just a mediocre one with unknown quality and an outdated, stale infrastructure. :) _______________________________________________ talk mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk
