Language Registrations needed for i-unknown and i-mixed
jon at hackcraft.net
Wed Jan 21 11:03:31 CET 2004
Quoting Bob Wyman <bob at wyman.us>:
> Addison Phillips wrote:
> >You should note that RFC3066 actually says the following
> > about the lovely 'mul' and 'und' codes:
> Of course, 'mul' and 'und' provide precisely what I was
> looking for in 'i-mixed' and 'i-unknown.' I should apologize
> for having troubled you folk with a request for something
> that was already there.
And I should apologise for mentioning mul and und again without checking for
more replies doing same. D'oh!
> I agree with what you say here. The problem is, of
> course, that RSS V2.0 has defined its own mechanism for
> conveying information about language and, as it is clear,
> that mechanism is much less flexible than that which is
> provided by XML itself. My fear is that there are a great
> number of home-grown parsers that process RSS feeds by
> searching for complete tags and those parsers will fail if I
> start inserting xml:lang attributes...
I wouldn't worry too much about parsers that look for complete tags, strive to
make your work interoperable but don't bend over backwards to cater to the
downright bizarre. However xml:lang doesn't (fully) solve your problem here,
xml:lang gives the language of the text in the feed itself, it is not
appropriate for giving the language of the item linked to.
You could always use RSS1.0 (just kidding, I do *not* want to introduce the 1.0
v 2.0 flame war to this list).
However, I think I recall someone doing work on using the Dublin Core namespace
with 2.0 along similar lines to how it is used in 1.0 - and it has an element
for describing the language of the resource in question. If I'm correct in that
then it might be an element that already has some adoption (particularly
amongst those aggregators and readers that handle both 2.0 and 1.0) so that
might be somethng to look for.
*Thought provoking quote goes here*
More information about the Ietf-languages