[Talk-us] Change sets covering entire US?

[Minh Nguyen]

Vào lúc 08:43 2021-11-08, Mateusz Konieczny via Talk-us đã viết:
> if someone would want to display names in English but in case of English 
> name being
> unavailable, show German name (which would be preferred over say "京市北")
> 
> In such case the solution would be to
> 
>      display name:en=*
>      if it is unavailable display name:de=*
>      if both are not present display name=*
> 
> But note what would happen if name=* would be in English, name:en=* 
> would be not
> tagged and name:de=* would be tagged:
> 
>      display name:en=* - skipped, as not present
>      if it is unavailable display name:de=* - done!
>      if both are not present display name=* - not reached

This isn't a theoretical concern either: modern operating systems and 
browsers allow users to specify a whole list of languages in order of 
preference. This preference is mainly used for language negotiation on 
the Web. [1] A user who speaks fluent English and also speaks Spanish as 
a heritage language may set both languages to avoid getting content in a 
third language on a foreign website. Nevertheless, they'd expect the map 
to label Philadelphia and Key West, not Filadelfia and Cayo Hueso.

> Alternative would be adding guesses about language of name=* (what is 
> tricky,
> complicated for many languages and areas impossible, fragile and 
> error-prone).

This decision is especially fraught in the U.S., which has no de jure 
national language, contrary to the default_language=en tag on the U.S. 
boundary relation. [2] In practice, English is a sound assumption, but a 
tribal community or ethnic enclave may have a distinct default language, 
and maybe only for POIs and buildings as opposed to streets.

Osmose has an issue filter "Default and local language name not the 
same" that compares name to the name:* indicated by the containing 
boundary's default_language. [3] Seen another way, this filter 
identifies "redundant" tagging of name:en. But I'm proud that fully two 
thirds of the Osmose level 1 issues attributed to me are due to this 
issue filter. I trigger it every time I fail to anglicize the name of a 
shop or place of worship that serves an ethnic minority. [4]

In general, it's still reasonable for a data consumer to assume name in 
the U.S. is in English unless otherwise specified, but it isn't so 
unreasonable to make that more explicit in the data, provided that the 
right processes are followed.

[1] https://www.w3.org/International/questions/qa-lang-priorities
[2] https://www.openstreetmap.org/relation/148838
[3] https://osmose.openstreetmap.fr/en/issues/open?item=5060&class=50601
[4] Or rather, when I go out of my way to OCR a sign in Korean or Arabic 
so I can tag it.

-- 
minh at nguyen.cincinnati.oh.us