Apr 20, 2023, 23:18 by ajt1...@gmail.com:

> On 20/04/2023 19:50, Mateusz Konieczny via talk wrote:
>
>> For start I want to propose to people to review shop tags in their area
>> with undocumented shop values or ones documented as problematic.
>>
>> See http://overpass-turbo.eu/s/1u2o
>>
>
> Reviewing "odd" tags before tagfiddling them away seems to me a very sensible 
> approach.  However, running that query locally finds, alongside a couple of 
> typos by me, lots that are very much correct, but just not on your list - 
> there are some very odd shops out there.
>

Note that I mentioned that many of them should be rather documented
(or new values covering them invented).

> To change "shop=veryrarevalue" where it was correct to 
> "shop=lessrarevalue"without preserving the detail somehow loses detail from 
> OSM and is therefore by definition a Bad Thing.  Some of the entries on your 
> list I'd definitely want to check onsite ("gun" and "firearms" is one obvious 
> one such, but there are others).
>
I agree in general, this batch is supposed to be listing 1:1 replacements
(I have similar list where I gather cases more suitable for shop=something + 
extra tag,
like shop=dog_groomer to shop=pet_groomer + pet=dog)

If you see any like this here, please let me know.

> Also, changing rare shop types into "yes" helps absolutely no-one.  If a data 
> consumer wants to handle a catch-all for "shop" they can; they don't need 
> them to be set to "yes" first.
>
It helps because maintaining lists of many many many rarely used meaningless 
values
in every single QA tool and validator and tool doing this is annoying at best.

For quite reasonable reasons JOSM developers do not want patches handling 
barely used
tags - if in 2025 someone uses say 50 instances of shop=needs_to_survey they do 
not
want to get a patch.

Also, there is no data loss whatsoever if confusing meaningless value (like 
shop=retail
or shop=??? gets changed into a standard meaningless value)

_______________________________________________
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Reply via email to