- From: Phillips, Addison <addison@lab126.com>
- Date: Mon, 18 Aug 2014 16:14:53 +0000
- To: Koji Ishii <kojiishi@gluesoft.co.jp>, Internationalization Working Group <public-i18n-core@w3.org>
> > 2) Isn=92t, or shouldn=92t, language determination primarily be made a > > user agent, and not a developer responsibility? > > I don=92t understand this question. Does the =93developer=94 here mean = > authors? If so, CSS WG thinks it=92s author=92s responsibility. The user agent cannot reliably determine the language of short runs of text and running a language detector is expensive. If we didn’t have @lang, there would be no way to get language-specific behavior for snippets of text quoted within a page or in multilingual documents (see for example the Wikipedia home page), other than inventing local styling and such. > > > 3) Does it matter at all? > > Yes, there are many cases where correct typography cannot be achieved = > without the content language, and @lang is the only tool we have today = for > authors to specify that. +1 Note that the "high runner" case for @lang is to mark the html element (e.g. <html lang=tlh-AQ dir=ltr>). I think the frustration here is that frequently language markup in pages is not present or is of distinctly poor quality. Better tools support would help. Better training would help. But throwing up our hands and taking away or deprecating the mechanism because many people abuse it, use it wrong, or are blissfully unaware of it does actual harm to those who do know what they're doing and rely on it. The more useful @lang is in producing good results, the better disseminated information about using it correctly will become. It was not that long ago that following this best practice actually did nothing at all in the display of Web pages. The situation is slowly but surely improving. When more support is integrated into the Web platform as a whole, I think the markup situation will also improve. Addison
Received on Monday, 18 August 2014 16:15:23 UTC