- From: Brian Behlendorf <brian@organic.com>
- Date: Thu, 9 Nov 1995 09:31:43 -0800 (PST)
- To: Maurizio Codogno <mau@beatles.cselt.stet.it>
- Cc: batie@aahz.jf.intel.com, www-talk@w3.org
On Thu, 9 Nov 1995, Maurizio Codogno wrote: > % Content Negotiation goes beyond browser type, but user type as well. For > % example, I would add this issue: > % > % 4. I want to send French pages to French speaking users, German, etc... > > Accept-Language: seems to me the best way to accomplish it. > My personal idea is to modify slightly a server, so that the resource > /foo/bar.html actually can correspond to either /foo/DE/bar.html or > /foo/FR/bar.html ... Apache uses bar.html.fr vs. bar.html.de, etc - in other words, configured a particular way, everything after the first . is considered a metadata keyword which maps to a namespace containing content-type, language, and encoding keywords. Thus, index.html.en and index.en.html would denote the same fact that this file is an English HTML file. Obviously this is not scalable - each of those dimensions should have its own namespace. So, we start the long and perilous journey into the metadata swamp.... something like GN/WN's .menu files, where the metadata is laid out explicitly and what's in the file name isn't important, is closer to what the right approach might be. But, these are all implementation details and largely irrelevant to HTML and HTTP protocols. Defining a standard for this is important when one considers cross-server portability important, which I do. Our work environment has to be focused on portability, because it simply isn't acceptible for us to force a client to use a particular server. On the other hand, is defining a convention for server-side data structure too much like reinventing CORBA? Brian --=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-- brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Received on Thursday, 9 November 1995 13:10:59 UTC