Bug#157734: request to support a wee bit of guesswork in content negotiation
When a user requests en-us, en-gb, en-au, en-ca or some other variant of en,
and Apache can't find a file in that variant of English, it doesn't try
simply en, but continues parsing the Accept-Language list.
% wget -q http://www.debian.org/ --header="Accept-Language: en-ca, fr" -O - | grep Language
<meta name="Language" content="French">
Despite the fact that:
% wget -q http://www.debian.org/ --header="Accept-Language: en, fr" -O - | grep Language
<meta name="Language" content="English">
In case the fallback (default) language of the server isn't up to the user's
liking, they will get an unknown language, and most of them won't realize
that their language preferences in the browser are at fault, that is, that
"English (<something>)" is not equal to "English (<any>)".
For the use in the Debian web pages (and 18 mirrors, a lot of which have
various default languages set up), we would really be grateful if a new
option could be added to Apache to make the daemon try the generic form of
the language first. Being able to restrict this behaviour to e.g. en-* would
be useful, too.
2. That which causes joy or happiness.