[Imports] Latvia-bot

[Mateusz Konieczny]

Jan 3, 2023, 07:29 by davisklavins at gmail.com:

> On Mon, 2 Jan 2023 at 22:32, Mateusz Konieczny via Imports <> imports at openstreetmap.org> > wrote:
>
>>
>>
>> Jan 2, 2023, 14:10 by >> davisklavins at gmail.com>> :
>>
>>> Thank you for the clarification! Updated everything as it makes sense to use one tag for all types of settlements and description of the tag matches the use case ("the name of the largest settlement (city / town / other) that is included in the address").
>>>
>> If you try to redefine tags then it should be clearly done, not hidden in a bot import proposal.
>>
>
> Not sure what you mean by redefining tags.
>
As I understand you started to used addr:place for cases where
there is street-indexed housenumber and no place-indexed housenumber
and it is not a conscription number.

Also, it seems that your bot is using add:housename for numeric values
that almost certainly are addr:housenumber

> In the State Address Register of Latvia, both numbers and names are within one field. In address notations, when there is no street, all numbers and names are put in double quotes as per > Articel 43.1. of the Regulation of the Cabinet of Ministers No. 455 of June 29, 2021 <https://likumi.lv/ta/id/324387#p43>> . Also note that in > Article 40 <https://likumi.lv/ta/id/324387#p40>>  on address notation in small villages without streets, no housenumbers are mentioned, only housenames. 
>
OK

> Thus, in Latvia, it makes sense to use addr:housename for all cases when there is no street
>
No. addr:housename is for house names (in this case tag is well named).
It is not for house numbers in any case at all.
Even if internal State Address Register of Latvia does not distinguish between names
and numbers, that does not change that OSM distinguishes between them.

For example https://www.openstreetmap.org/way/237853362 and
https://www.openstreetmap.org/node/9888037895 have addr:housename=1is at least extremely suspicious.

And if you mention that "both numbers and names are within one field" then it needs
to split into proper fields during import.

Rather than assume that OSM tagging follows Latvian legislation and 
redefining addr:housename to "house names and whatever
is stored in Latvian state databases in the same field as house names".

> Read my previous answers and rethink again your ideas about rolling back.
>
I am still pretty sure that addr:housename=1 is a clear tagging mistake.

"In the State Address Register of Latvia, both numbers and names are within one field"
does not mean that they should or can be stored in the same OSM tag.

(I would be more likely to classify it as "weird local traditions" if that would be discussed
before edit and actually planned, now it looks like "it is feature not a bug")
(I am not planning to make any rollbacks personally, but I expect that it will continue
to break processing of OSM data in Latvia and sooner or later it will be fixed)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20230103/1b7d8740/attachment-0001.htm>