- From: Gary Adams - Sun Microsystems Labs BOS <gra@zeppo.East.Sun.COM>
- Date: Tue, 14 Jan 1997 10:20:29 -0500
- To: "M.T. Carrasco Benitez" <carrasco@innet.lu>
- Cc: WInter <www-international@www10.w3.org>
M. T. Carrasco Benitez writes: > Should it be defined a data structure to keep the linguistics versions in > the server side. > > Tomas > Is this question about how to standardize on a common implementation for servers addressing multiple variants of the same document. In the past people have used a filename extension, an alternate directory path, an alternate server/port to identify variants. All of these solutions were based on "addressing" in that they changed the URL. e.g fromage vs. cheese The content negotiated approach leaves the name space intact, but uses attributes (parameters) of the request to select the appropriate variant. e.g. cheese(swiss) vs. cheese(cheddar) Since some servers are built on top of filesystems and other servers are built on top of databases. I don't think a "data structure" would be the appropriate form for registering the linguistic versions. Recently, I began using the Jeeves http server. In that environment subclassing the basic http request to include Language support (similar to what has already been done with MIME support) would be a natural extension of the current Java classes. (Don't take my bias as hype, but as an example of something that is neither a file system or a database that will need to comply with RFC2070 in HTML 3.X ) http://www.javasoft.com/products/java-server/alpha2/doc/architecture/ http://www.javasoft.com/products/java-server/alpha2/doc/apidoc/sun.server.http.HttpServletContext.html
Received on Tuesday, 14 January 1997 10:24:05 UTC