Re: ISSUE-88 / Re: what's the language of a document ?>,,, 
Hi, Ian, Leif, all.


I like Leif's solution--to use the first language specified in http as the text processing language when none is specified in the html tag.

Regarding the desire to deprecate the specification of multiple languages in the http header, I thought it was the W3's job to create standards that would facilitate identification of content in other languages, including content in multiple languages on the web, at some point in the future, and not to 
restate what the current vendors' products are capable of.

However, for my part I still assume that one possible use of http-equiv today would be when
I had a page in two languages targeting speakers of both 
(even  though you all tend to prefer that such a page be divided into two pages for various reasons that is not always done), then content-negotiation could pull up the page for speakers of both languages.

The rest of the stuff I mentioned--having search engines select content in Old French with modern French summaries for example--
might be best left to the meta data (keywords, description Content-Language in the meta element)
where it would be available for search engines -- when they get to a point where they can retrieve a page in multiple languages according to someone's preferences to have some content in one but summaries or explanations in another.

But at the same time when the page is truely in two languages to the point that speakers of both might be able to read enough of it, I'd like the http to be able to specify both languages too.

(Sorry I have not got Roy's technical expertise; I appreciate all he did to explain the uses of http today but can't be as technical.)

From: Leif Halvard Silli <> 
Date: Thu, 11 Mar 2010 08:57:09 +0100

> Ian Hickson, Thu, 11 Mar 2010 07:39:08 +0000 (UTC):
>> On Thu, 11 Mar 2010, "Martin J. Dürst" wrote:

>> When a feature is for use on a server by a server, interoperability isn't 
>> needed and therefore the HTML5 spec is irrelevant. Servers are free to use 
>> whatever mechanisms and conventions they want to get whatever effects they 
>> want on their end.
>>> As a result, the HTML5 spec best should just say that <meta http-equiv 
>>> is used primarily as meta-information on the server side and is 
>>> therefore in general ignored on the client side.
>> It's not ignored on the client side in practice.

>Yes it is. As long as you use the @lang attribute, then META 
>content-langauge has no effect. Hence servers should be free to use it 
>as they want = according to the HTTP specs.

> So, hereby I propose a compromise solution:

> If the HTML document *doesn't* use the @lang attribute on the root 
> element, then the content-language pragma is forbidden from containing 
> more than one language tag - and this language tag will also define the 
> language of the document.

> However, if the document does use the @lang attribute on the root 
> element, then authors are free to use 'http-equiv="Content-Language"' 
> for what it is meant for according to HTTP.
I like this solution.

From: Ian Hickson <> 
Date: Thu, 11 Mar 2010 04:39:06 +0000 (UTC)

On Wed, 10 Mar 2010, CE Whitehead wrote:
>> my reason for specifying multiple languages is that the content is split 
>> between the two and that to read it you need both languags to at least 
>> some degree; so for when search engines and such decide to pay attention 
>> to http it will be useful.

> It seems to me that for that use case, the lang="" attribute and language 
> recognition heuristics be a better indication than the Content-Language 
> pragma. I don't think that argues for the pragma at all.
Well as we've been discussing, lang="" does not allow two languages to be specified; what about the case when a page is in one language with summaries in another?
Is there any reason not to let it be recognized as targeting both audiences--that is
to only specify the language of  one of the two audiences?
C. E. Whitehead
> -- 
> Ian Hickson               U+1047E                )\._.,--....,'``.    fL
>       U+263A                /,   _.. \   _\  ;`._ ,.
> Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'


Received on Friday, 12 March 2010 01:03:16 UTC