RE: ISSUE-88: content-language-multiple - Chairs Solicit Proposals

Richard Ishida, Wed, 9 Dec 2009 09:05:01 -0000:
> Hello Leif,
> 
> I was told that in your change proposal you will propose that the lang
> attribute supports multiple language values.

Whatever the reason: the one/ones who told you so was/were in error. 

>  Although we haven't discussed
> it for this particular topic, I'm certain that the i18n WG will strongly
> oppose such a suggestion, based on a long history of working with and
> educating about the language attributes.  The main reason for this is that
> the language attribute defines the language of a range of text for
> text-processing purposes, which requires  information about a single
> language - use of multiple language values makes no sense for that.  Another
> reason is that for consistency similar changes would have to be made for
> xml:lang - and the likelihood of that happening in the near timeframe is
> essentially zero.
> 
> For more information about this and the difference between values of
> language attributes and those of the HTTP header or the meta
> Content-Language element, see
> http://www.w3.org/TR/i18n-html-tech-lang/#ri20040808.100519373
> 
> Could you please confirm to us whether you were planning to propose the
> above, and if so could we please discuss this (and indeed any other
> divergences from the proposal made by the i18n WG at
> http://lists.w3.org/Archives/Public/public-html/2009Oct/1086.html) before
> you submit your change proposal?

I have followed the debate fairly thoroughly, and I also filed the bug 
report that lead to Issue-88. I gave my interpretation of the I18N WG's 
proposal here:

http://www.w3.org/mid/20091126202756258786.e7d3d2a4@xn--mlform-iua.no

Perhaps be the confusing point in that letter were the following:

]]
But since the meta may also be used to set the 
language of the document, if the lang attribute is lacking or wrongly 
set in the <html> element - or consciously do so, then - for that 
purpose - if you place multiple languages inside the meta element, then 
it is equal to setting multiple languages inside the lang attribute. 
[[

When I said the above, I meant exactly the same that you expressed in 
the I18N WG's proposal message which you pointed to above:

]]
[4] Establish the rule that multiple values in the place that has 
precedence
equates to lang="".
[[

That is, if someone defines the audience languages like this: 

<meta http-equiv=content-language content="en" />

then it may also be interpreted as setting the document language to 
"en":

<html lang="en">

Whereas if sets the audience languages to these:

<meta http-equiv=content-language content="en, de, ru" />

then it must be interpreted as if the document language is unknown:

<html lang="" >

Please let me know if my interpretation deviates much from yours.

By the way: I offered to write the change proposal because I filed the 
bug - I felt that was like taking responsibility  for ones acts. 
However, I would be happy to step down from that duty, and have been 
considering that thought since you entered the thread asking about the 
next step. After all, I read that you talked with Ian at the W3C 
conference recently and so on - it sounded as if you were approaching 
and understanding. Please let me know if you think that would be in 
order.

BR
Leif Halvard Silli

Received on Wednesday, 9 December 2009 20:03:16 UTC