- From: Chris Lilley <chris@w3.org>
- Date: Fri, 9 May 2003 21:50:18 +0200
- To: jaccoud@petrobras.com.br
- CC: Dan Connolly <connolly@w3.org>, Ivan Herman <ivan@w3.org>, w3t@w3.org, w3c-translators@w3.org
On Friday, May 9, 2003, 8:58:40 PM, jaccoud wrote: >>Hmm... that's an interesting idea... what sort of query did >>you have in mind? jpcb> The trick is restricting the check in xml:lang attribute to just jpcb> the number of characters present in the query. This way, a query jpcb> to "pt" would return any documents tagged with either "pt", jpcb> "pt-BR", or "pt-anything". If you go a step further and restrict jpcb> the test to the minimum number of charaters of either the query jpcb> string and the testes string, then it would also return "pt" jpcb> documents when asked for "pt-PT", but not "pt-BR". Right. And further, it has to be seen as a hyphen-separated list of tokens. Its not a substring match. So "p" or "pt-b" should not match anything. jpcb> I think this jpcb> behaviour is exactly what is expected from xml:lang. Yes. jpcb> This is jpcb> easy to implement, just tweak the XPath expression in the query jpcb> using the string-length() and starts-with() functions. Possibly. -- Chris mailto:chris@w3.org
Received on Friday, 9 May 2003 22:49:12 UTC