Peter Körner wrote:
Maarten Deen schrieb: They tell you that the translation in the given language is identical to the value of the name=* tag. If you see the name-Tag as a fallback for a missing name:xx-tag (what you should), those pseudo-translations are needless. I'm currently in a discussion with Marc Schütz (search through the mails of the last days) if deleting them is a good or a bad thing.
I have read the above mentioned discussion[1] (unfortunately it is spread among two mailing lists) and I have two additional points to make:
(1) Default tags can be changed. We should remember that default tags can be edited by somebody later and they will no longer be good for other languages.
(2) There is some inconsistency in default tags. Sometimes it's the English name, sometimes it's written in the Latin alphabet, local alphabet (e.g. Arabic) or both. I think Iran is spelt in Arabic, Comoros are spelt in both. Some people say Burma, some say Myanmar for various reasons. I think having explicit name:xx tags even if *at the given moment of time* it's the same as the default.
That's said, I have added "name:pl" to "Polska" even thought it is the default name, too. Therefore having "name:de" == "Deutschland" is perfectly fine. In this case it actually indicates that the local official language is Hochdeutsch ("de" or "de_DE").
Therefore I would propose to remove "orange" tags from the utility - such name will be either "italic" or "orange" and never "normal". Both carry notion of something being wrong with the name.
I actually wonder if the default tag is the right thing to have altogether. Probably better might be to use some fallback order (say, "en,de,ru" to be very European-centric) and displaying the name in italic in OSM (meaning "fallback language applied").
Some more intelligent fallback mechanism could be applied in the future (using user's browser preferences for example):
- Browser says "Accept-Language: zh;q=1.0, ja;q=0.2, en;q=0.1" - this means "I understand Chinese (say Mandarin) and a bit of Japanese and some tiny English". For more details on that see RFC2616[2].
- The webserver sees that there is no "name:zh" but there are "name:en" and "name:ja". This user indicates it prefers Japanese to English. Actually in this case Japanese is much better option for the users since there are chances that the kanji spelling will be the same as Chinese, like, for example, 中国 (same in the Japan language and simplified notation of Chinese).
But this would require on-demand application of the negotiated labels on the map and this technically might not be easy to be done in a feasible name (it would be difficult to create pre-generated tiles for different sets of user preferences).
[1]http://thread.gmane.org/gmane.comp.gis.openstreetmap/39745/focus=39900 [2]http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html