I am a recent mapper and would appreciate the opinion of more experienced users regarding Gaeltacht Placenames.
Should we do away with the English version of placenames within Gaeltacht boundaries?
English placesnames in the Gaeltacht, e.g Muckenaghedderdahaulia, Lough Natawnymore, Illaunnagappul, are not recognised by the state, said by local people or printed/spoken in media.
“The Regulations provide that the Irish language version only of a Gaeltacht
placename – as declared in any Placenames Order under part 5 of the Official
Languages Act 2003 – may be used on the maps specified in the Regulations (i.e.
large-scale definitive maps prepared and published by or with the permission of
Ordnance Survey Ireland on or after 1 January 2005).”
It’d be great to link to the OSM objects - I’m guessing that your first one might be here. There name is in Irish (as you’d expect), and there’s also an English version and a couple of alt_name in both languages.
Is the situation that there is “not an English name at all” or that “the OSM English name is really an old name and not used by anyone; in reality the name used by people speaking English is actually the Irish name”?
Disclaimer - I’m English, but do create maps in various languages that include Ireland; the languages include Irish, English and (for Ireland) “whatever is in the name field”.
Just keep the English name tagged as such, just the same as you’d do if adding a name in any other language. (Look at Galway for instance - it has name tag variants in all manner of languages Relation: Galway (1390623) | OpenStreetMap )
I don’t see that the fact a place is in a Gaeltacht area has any bearing on the matter. The Law you cite applies to ‘definitive maps’ published by Ordnance Survey Ireland, it has no bearing on OSM.
Which layer are you looking at? The default (“Standard”) layer shows the name tag, not any of the name: language-specific tags - so you should generally only see Irish names in the Gaeltacht. Other layers may work differently.
The vector map on osm.org uses whatever language is configured in your browser. I just tested setting my browser to Gaeilge and the name that’s displayed by the vector map changed accordingly.
Thanks, indeed other layers display the name tag of the preferred language.
Although this is my point that many Gaeltacht placenames are unwieldy in English and not in use at all but taken from historical maps. Should the decision be made on a case by case basis?
Do you mean English-speaking people are using the Gaelic names?
Then maybe move the name:en to old_name:en and copy the Galeic name into name:en.
IMHO, I doubt English speaker use the diacritic characters (accents…), but I’m not local.
English speakers may pronounce the placenames in an anglicised way but they wouldn’t spell some of them in English, because of their unwieldiness.
I do understand the use of name:en for placenames in the Gaeltacht. But I do find it a shame in many online services I use, that use OSM, the map-tiles render name:en for Gaeltacht regions, placenames that look and sound bizarre to the people living there. Of course these services aren’t offered in Irish, so the service is rendering a map to the user’s preference langauge (i.e English). I do wish there was some exception to certain cumebrsome English Gaeltacht placenames.
I also find it odd that some maps do that. For Wales I went with name:cy for the most Welsh speaking areas and name:gd In northwest Scotland. Ireland generally has sensible values in name, so I just used that as the default (though the vector maps have a language switcher to switch between the 5 choices).
There’s a discussion happening at the moment for Scot’s Gaelic. name: has been set bilingually, e.g Tobermory ~ Tobar Mhoire and Budhmor - Bugha Mòr. Portree was recently changed back to English name and as seen in the discussion there’s not an easy way of filling both boots.
Thankfully name: in Gaeltacht areas will remain as the Irish language placename but I would suggest moving outdated, unwieldy Anglicised versions of these places to old_name:en.
Yikes. Dual names with a common separator are bad enough for data consumers to make sense of, but dual names with an arbitrary little-used separator such as “~” are actively sociopathic.
If people must do this to work around (yet another) osm-carto failing, please just use a “/” or “-” separator like everyone else?
In principle I think that’s reasonable, if no English speakers use those versions when speaking or writing in English.
In practice, the difficulty may be in establishing which English names have fallen out of use. I think that for some better-known Gaeltacht locations like Carraroe or Spiddal we are not yet at the point where we should move the English name to old_name:en(although it might well happen in the future). For lesser-known places it might be justifiable, but I’m not sure how you would make that distinction.
I don’t see what this has to do with Carto. It’s a raster map, it can render one name, it renders what is in name. The problem here seemed to be some renderers always preferring name:en?
Vector maps can be more flexible.
Either way there’s no excuse for jamming everything in name.
An example of “what used to be an English name no longer is” can be found across the water at Porthmadog. Since 1972 the Welsh variant of the town name has been both the official English name and regularly used by most people as the name in English.
Perhaps the same logic applies to a lot of the Galway Gaeltacht (although I suspect not to e.g. Spiddal / An Spidéal)?
Oh, it’s simple really. Each language pair or region gets to pick its own special delimiter. That way, a user can easily tell which language pair or region they’re looking at, just from the map label itself. A handy decoder ring is available somewhere in the annals of the relevant talk-xy mailing list. Best of all, it isn’t a semicolon, so you know you’re looking at a name=* rather than an alt_name=*.
This coded approach is inspired by our globally coordinated system of alphabetic prefixes in way refs. Thanks to this system, some data consumers have gone years without needing to handle route relations. A decoder ring is available on the wiki, with a warning at the top that it’s controversial.
As for the dual names, Unicode has plenty of punctuation characters, enough for everyone to pick their favorite. But no emoji please. That’s reserved for future use.
(With apologies to anyone with a tilde in their name.)
tl;dr - for every other tag in OSM, multiple values are separated by a semi-colon (for example, name=York;Efrog) and have been since 200mumble.
osm-carto refuses to add support for this, because of the “Han unification problem” or something. (I am not making this up. It’s literally mentioned in the linked ticket as “the other big problem of multilingual names”.)
Consequently mappers, in a misguided attempt to make osm-carto look less like ass, tend to use slashes or dashes (or, in this case, tildes) as separators, because “York - Efrog” looks less terrible than “York;Efrog”. Most maps would substitute in a newline or just choose the first or something, but Carto gonna Carto.
Just because you can put multiple things in name doesn’t make it a good idea. We don’t generally support things like amenity=cafe;doctors. If name=X;Y is pretty-printed as “X / Y”, then this would be indistinguishable from X / Y for name:left / name:right (which would be a more useful convention to support IMHO).
I find it a fair choice not to pretty print and to encourage mappers to clarify semantic differences.
The decision of whether to pretty-print is up to the individual style’s designer and the reason that we “don’t tag for the renderer”.
Not enough renderers understand name:left/right=* or are even capable of boundary edge labels, but of those that do, most effectively use whitespace as the delimiter. That is, they label the left name on the left side and the right name on the right, as OSM Carto does, or even flip one name upside down.
If a renderer does choose the slash or dash as a delimiter for stylistic reasons, it does so knowing that a single name in a single language can also contain a slash or dash, which should never be interpreted as a delimiter. A semicolon on the other hand is intended to be pretty-printed.
That said, if the second name in name=* is superfluous or contrived, then by all means, don’t force trivia on data consumers.
