W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > April to June 1997

Re: Open Requirements Issues

From: Larry Masinter <masinter@parc.xerox.com>
Date: Fri, 6 Jun 1997 12:44:44 PDT
Message-ID: <339868AC.50F9@parc.xerox.com>
To: "Gregory J. Woodhouse" <gjw@wnetc.com>
CC: webdav@warlok.ds.boeing.com, w3c-dist-auth@w3.org
The proper label for the language of content is
content-language:, if it is part of the wrapper. Some
media types (e.g., text/html) also have provision for
labelling components as being in a particular language.
However, there isn't a direct correspondence between
accept-language (which describes the preference of
the user of the client) and content-language (which
denotes the language of the content for purposes of
viewing, hyphenation, display conventions, etc.).
> Similarly, a PUT (or check in)  should be able to specify a language (It's
> a little non-intuitive to use "Accept-Language" here, but I suppose it
> would work. If the server is unable to store language specific versions,
> or is unable to store the document under the language specified, I believe
> it should respond with

I think PUT on negotiated resources for which the variants don't
have URLs should basically be disallowed. (Well, you might PUT a
multipart/alternative and then expect the result to be content
negotiated as a response). That is, if you have
which is really 
   foo.fr.html and foo.en-us.html
and you just want to update the French version, then you should
PUT to foo.fr.html and not use some arcane invocation of PUT foo.html
which says "only replace a part of this".

It's probably the same reason why PUT on byte ranges should be

Received on Friday, 6 June 1997 15:45:41 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:01:15 UTC