- From: Ernest Cline <ernestcline@mindspring.com>
- Date: Wed, 24 Sep 2003 14:13:26 -0400
- To: www-html@w3.org
> From: Reuven Nisser <rnisser@ofek-liyladenu.org.il> > > Not to define the language at all is a possibility but I fill that it's like > throwing the water and the baby. I agree that defining Content-Language META > with the list of languages is enough for the W3C standard. But still W3C > standard needs to define exactly which language to be use for each character > so we still need to define the rules for that. To expect user agents to make such distinctions seems to me to be asking them to make too many assumptions given that there is no one-to-one correspondence between script and language. For instance, there is the example given earlier in this thread about numbers in a mixed English-Hebrew document. There is no simple common sense rule that a user agent could be given. As a result, I don't think that such an ability should be added to HTML. At best, perhaps doing something like the following to indicate ranges for languages could be considered for XHTML2. <html> <head> <charlang value="he:"U+0030-U+0039,U+0590-U+05FF;en:U+0021-U+007E" /> But I don't like it. and the more flexible alterative of adding a charlang attribute to everything I like even less. Trying to shoehorn this into the existing lang and xml:lang attributes has the problem that existing implementations won't understand the new values, so that is clearly not acceptable at all. (Note: The intention is to indicate character ranges using the same syntax as that used in CSS2 for character ranges in fonts, and that a character is bound to the first language encountered, thus eliminating the need for breaking up the range for the English characters into two discrete ranges and providing a defined behavior for when such double definition occurs.)
Received on Wednesday, 24 September 2003 14:13:38 UTC