Mangled ref:name from import

Communities Ireland
1

I came across some invalid data in OSM, and while trying to investigate that came across some Irish relations https://www.openstreetmap.org/relation/14571417/history and https://www.openstreetmap.org/relation/14574362. In both cases it looks like what is now ref:name hasn’t been properly converted from UTF-8 at import**.

I presume that in both cases ref:name should be the correct (Irish) name of the town?

Edit:
Also https://www.openstreetmap.org/way/1093798199/history, https://www.openstreetmap.org/way/1094051558/history

** in both cases the original adder of the data was someone working for Microsoft - insert your own “Micros~1” joke here. :smiley:

Debugging UTF-8 errors with OSM .pbf files
Debugging UTF-8 errors with OSM .pbf files
2

Everyone has an Irish relation Andy, wouldn’t you know from Italia ‘90 yourself? :grin:

Maybe @risteárd_ó_ha would be well placed to look at this as he was doing some great work around Irish names in Gaeltacht areas.

1 Like
3

Indeed both these towns are in the Gaeltacht so ref:name should equal name:ga (which should also be name).

I’ve experienced issues with diacritics importing CSVs to Microsoft Excel, which didn’t exist with Google Sheets, Open Office, MAcOS Numbers etc.

1 Like
4

I’ve updated ref:name for both. I’ve left name as is in case there’s anything esoteric about “names being added to relations as well as nodes” - someone can always add it if necessary.

1 Like
5

Census 2015 Boundaries or more properly late data releases from Census 2011. Hmmm.

These are synthetic town boundaries and are very valuable data as they show the real extent of towns and villages in Ireland, there are no legal town boundaries and even were there so there is no space in the admin-level= stack for them.

My problem is that these should have been superceded by more recent Census 2022 Boundaries, reflecting recent changes, and arguably the 2015 boundaries should then be deleted as irrelevant.

There is a case for using OHM compliant tagging on ephemera like this.

Start_Date=2011 and End_Date=2026 would be enough to satisfy OHM and the old boundaries can be ingested to OHM prior to deletion and import of current boundaries. That way old towns live on where they belong and OSM is not unduly cluttered.

We completely missed the 2016 settlement boundaries referred to in this link below. 20 ‘new towns’ were created. This link has the ‘zen’ article explaining where settlement boundaries come from.

Census 2022 town/village boundaries are most current and were released in 2023. The nomenclature has changed to a “BUA” which might challenge OSM tagging ..a tad.

Thoughts?? My feeling is that OSM should either have current boundaries…..or none.

6

Are you sure there’s no de facto ‘town boundaries’? In Northern Ireland, the planning service at each district council defines ‘development limits’ for each settlement, which are effectively the town and village boundaries. (They’re generally published in 10 year district plans).

Is there no equivalent thing in the south?

7

Closest thing in the south is the Census Settlements which are a 5 yearly Census output.

Town boundaries were a thing up to 2014 but while they were legal they were chronically out of date. Town councils were abolished in 2014 as was Limerick City in 2019.

So Census Settlements or “Built Up Areas” are the closest we have.

8

Some councils do specify development limits or equivalents. Those are only for town planning purposes though and not for other administrative purposes. However, in Northern Ireland the difference between urban and rural can often be stark (a chunk of it down to the things the Housing Executive had to do and why it had to do them), whereas in the Republic towns and boundaries are rather amorphous.

Within their own administrative logic, this affects things like speed limits, and the presence / absence of footways and street lighting.

1 Like