W3C home > Mailing lists > Public > www-international@w3.org > April to June 2010

RE: Null change proposal for ISSUE-88 (mark II)

From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Date: Tue, 6 Apr 2010 14:44:06 +0200
To: CE Whitehead <cewcathar@hotmail.com>
Cc: ian@hixie.ch, public-html@w3.org, julian.reschke@gmx.de, www-international@w3.org
Message-ID: <20100406144406924189.d5b0d246@xn--mlform-iua.no>
CE Whitehead, Mon, 5 Apr 2010 23:40:11 -0400:
 ...
> So I cannot go with Leif's suggestion that you go all the way and 
> give all priority to the http header.

Not a suggestion. Merely an assertion that it would have been *logical* 
if http header took precedence. However, as since noted: what HTML4 
says about the "default-style" header/http-equiv, both seems more 
logical and more factual: [1] "If two or more META declarations or HTTP 
headers specify the preferred style sheet, the last one takes 
precedence." 

Note that in Ian's zero change proposal, then every time the META c-l 
contains a comma separated list or is empty then, quote: [2] "language 
information from a higher-level protocol (such as HTTP), if any, must 
be used as the final fallback language instead." (Bug 9420 [3]) 

In today's user agents, the last META c-l is always the end station. 
(Except in Mozilla, whenever the value of the META is the empty string 
{bug 9422 [4]} - here Mozilla differs from how it treats the META 
default-style declaration.) Thus Ian's zero change proposal will make 
user agents pay many more visits to the http header than they do today.

If HTML5 took the full step and made the processing of META 
content-language and lang="*"/xml:lang="*" 100% identical, then the we 
would be closer to reality. (Bug 9417 [5])

> Thus I remain in favor of the status quo basically:  I think the 
> html4 specs are fine.  I think for inheritance of the text processing 
> language, first priority should be to the html lang=  followed by the 
> last meta content-language element in cases where there is more than 
> one (here I go with Leif).

Indeed. This is where I went. ;-)

> I favor that the meta content-language element take a comma separated 
> list of one or more languages, or alternately the empty string 
> (null).

+1 See my last letter. [6]  Basically: 

Syntax = allow comma separated list. Processing = depends. If vendors 
do not agree with Mozilla, then processing should be identical with 
lang/xml:lang. This is also status quo. (Currently there is no such 
agreement, thus this is also what I suggest.) If vendors do agree with 
Mozilla, then we must define the exact algorithm (bug 9426 [7]). But 
regardless: the empty string must have the same meaning that it has in 
lang/xml:lang.

>  Same for the http header which should still get priority for 
> content-negotiation which I cannot benefit from.

[1] http://www.w3.org/TR/html4/present/styles#h-14.2.1
[2] http://dev.w3.org/html5/spec/dom#the-lang-and-xml:lang-attributes
[3] http://www.w3.org/Bugs/Public/show_bug.cgi?id=9420
[4] http://www.w3.org/Bugs/Public/show_bug.cgi?id=9422
[5] http://www.w3.org/Bugs/Public/show_bug.cgi?id=9417
[6] http://lists.w3.org/Archives/Public/public-html/2010Apr/0146
[7] http://www.w3.org/Bugs/Public/show_bug.cgi?id=9426
-- 
leif halvard silli
Received on Tuesday, 6 April 2010 12:44:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 6 April 2010 12:44:42 GMT