- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Thu, 17 Apr 2008 12:48:43 +0300
- To: HTML WG <public-html@w3.org>
It seems that some authoring tools and authors use <meta http- equiv='content-language' content='languagetag'> instead of <html lang='languagetag'>. http://philip.html5.org/data/meta-http-equiv.txt Based on the usage pattern, I think authors mean to use <meta http- equiv='content-language' content='languagetag'> in a way analogous to <base href='uri'>. That is, as a declaration that belongs between HTTP and the root element in the inheritance chain based on an obvious guess about author intent. Moreover, with FrontPage, this isn't invisible metadata, because a faulty meta content-language is visible to the author as squiggly red spell checker lines. The spec should probably say something about this. Issues: * Would observing content-language meta Break the Web? Currently Firefox 3 beta 5 doesn't observe meta content-language in font selection. Demo baseline (verify that "No lang: 中" and "zh-CN: 中" show up in a different font): http://hsivonen.iki.fi/test/lang.htm8 Content-language demo (compare font with previous): http://hsivonen.iki.fi/test/moz/content-language.html * Would observing HTTP content-language Break the Web? That is, how common is it that HTTP servers send bogus content-language picked up from the system locale, for example. It's quite usual that the server runs as en-US, because the admin wants to maximize the googlability of error messages and the applicability of documentation, but the content served is something else. Even you en and en-US were flagged as untrusted values, it happens that multinational system integrators source server setup to another country within Europe, so the person who set up a server for Finnish use may have installed the system with setting suitable for a Danish keyboard layout for example. * The de jure definition of HTTP content-language isn't exactly suitable for the lang inheritance chain. However, authors seem to use content-language in a lang-like way. * What to do when content-language comes with more than one language tag? * Will only the first occurrence in "*the* head element" count as with base? * Will a tag without the content attribute count as the first occurrence? If so, will content default to "" (unknown)? -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/
Received on Thursday, 17 April 2008 09:49:30 UTC