W3C home > Mailing lists > Public > www-international@w3.org > January to March 2010

Re: ISSUE-88 / Re: what's the language of a document ?

From: Roy T. Fielding <fielding@gbiv.com>
Date: Thu, 11 Mar 2010 10:40:32 -0800
Cc: Richard Ishida <ishida@w3.org>, www-international@w3.org, public-html@w3.org, Maciej Stachowiak <mjs@apple.com>, Ian Hickson <ian@hixie.ch>
Message-Id: <A84B71E9-E0DA-4E32-9CED-DF8516455B1F@gbiv.com>
To: Henri Sivonen <hsivonen@iki.fi>
On Mar 11, 2010, at 3:51 AM, Henri Sivonen wrote:

> "Roy T. Fielding" <fielding@gbiv.com> wrote:
> 
>> That is a completely false statement on all accounts.  First, it is
>> obviously data about the data, and thus metadata.  Second, it isn't
>> particularly intended for the user agent (the primary purpose for it
>> is for the author to help the server infrastructure understand the
>> audience of this content and thereby influence content negotiation).
>> meta http-equiv is a part of HTML specifically designed to assist
>> with content management.
> 
> Why is Content-Language delivered to the UA if it is meant for the server's use?

Because the UA is supposed to be a hypertext editor, not just
a browser.  And, in some cases, it is.  This data is something
that the author needs direct control over.  Hence, it is
embedded in HTML and defined by MIME/HTTP semantics.

> Are there content management systems that ingest documents from another server and use Content-Language to decide how they re-serve the document?

Yes, though it is far more common for CMS systems to ingest that
information on import from some legacy platform in order to build
a multilingual or multisite system, e.g.

http://dev.day.com/docs/en/cq/current/administering/multi_site_manager.html

The typical use case for content-language is to supply the data to
external processes that periodically parse the content and
prepopulate some other database (e.g., Apache .var files, CERN meta,
WN configuration, JCR properties, etc.).  This stuff goes on in
the background.

The only role http-equiv has in this is to provide a metadata
namespace that all of these systems agree to as a standard.  HTML5
has no right to change the meaning or control the values within
that namespace, nor is there any need to do so.  Everything that
Ian wants to accomplish regarding specifying browser behavior wrt
content-language can be done in a separate section on how to determine
the language.  That section can easily restrict the default to a
single language without impacting the definition of external
standard metadata.

The docs at

http://www.w3.org/International/questions/qa-http-and-lang

are correct.  It would be best if HTML5 didn't break them
and all of the systems that depend on those semantics being
part of the Web architecture.

....Roy
Received on Thursday, 11 March 2010 18:41:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 11 March 2010 18:41:09 GMT