Hi! We've talked a lot about that when we discussed Zope I18N some time ago. Most of this is relatively easy to accomplish even now. Regarding language negotiation: Both Localizer and ZBabel do this. The problem in my practical experience is that most people do not configure their browsers correctly. This is especially true with people having English instead of localized browser versions (e.g. in large corporations), but at the same time not expecting to get The Web in English only ... -- I hope that this will change ... With regard to content type: I had a proposal some time ago that I still think to be cool, but that didn't get too much attention: If we used object names without any .whatever suffixes, we could easily provide handlers for different output formats. An example: Let's say you upload an OpenOffice text file in XML format. It is called "myFile.sxw". It will actually be stored as a folderish object called "myFile" that contains all the items of an OpenOffice file (which is a zipped folder of content, stylesheet, and image files). You could then retrieve it as HTML, using myFile.sxw, as a MS Word document, using myFile.doc, etc. etc. This might not be the right thing as a Zope default. But in a document management context this would just be marvelous. The decision what file format to return could be handled via content type negotiation, or via an explizit URL containing the suffix. The only thing that still is not quite obvious to me is how to handle FTP or WebDAV requests. One could display ALL options, or just the default one.

I try to keep tabs on other open source CMSs, in particular Apache Cocoon, and I noticed this message:

http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=103717736917834&w=2

talking about software that, in effect, enables one to configure a server to respond to the HTTP headers having to do with content negotiation: preferred charset, preferred language, and content-type.

This is way cool stuff-- imagine an application capable of returning an XML representation of a resource instead of a human-readable XHTML simply based on the HTTP header-- same URL.

ASIDE: This is in line with TBL's original vision of the web and is being adopted by high profile organizations like google-- it has supported Accept-Language for at least a year-- change the list of acceptable languages to demonstrate it in a browser...

Question: has anyone tried to support this kind of thing using Zope yet? I imagine it should be possible, b/c you should have access to everything you need in the REQUEST object, right?

If one agrees that this is a good idea, then wouldn't it be cool to have a set of ready-made components in Zope3 to facilitate switching on relevant HTTP headers-- either for mime type, I18N, etc.?

Thoughts?

--Craeg

_______________________________________________ Zope-Dev maillist - Zope-Dev@zope.org http://lists.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - http://lists.zope.org/mailman/listinfo/zope-announce http://lists.zope.org/mailman/listinfo/zope )