[standards-jig] I18N (Internationalization)
max at quendi.de
Thu Mar 14 16:00:17 UTC 2002
First, let me tell you why I think I18N is important: there are many
jabber components that have to interact with users. Some examples
include iq:register, iq:search; iq:gateway, and soon maybe x:data.
These all have to contain some text, for instructions, field labels
etc. Currently, all of these are usually in english. This is OK while
Jabber is just a geek thing, but to target the broad masses, I18N of
so central things as searching for buddies is crucial for the
survival of Jabber. Hence we have to support it, and we better start
work on this now than later (actually, it is already quite late,
IMHO, considering we talked about this 6 months ago).
I browsed through my private list archive. and found that Julian and
I discussed this topic here before. So, here's an old message from
Julian, and a (new, different from the orignal) reply by me. If I
missed some development in this area in the past months, please
excuse me and point me to a place where I can get additional
At 14:06 Uhr -0500 04.11.2001, Julian Missig wrote:
>Well, I suggested a *long* time ago that the initial stream:stream
>should carry an xml:lang attribute, which the server can then
>identify the language of a session with (and if we want to get more
>complicated, we can allow xml:lang anywhere in the session - if they
>do an iq request with an xml:lang different from the session
>xml:lang, the response should be in the same xml:lang... but that
>might be a bit much).
I think it's important to allow clients to tag xml:lang on any
message/presence/iq for now, at least as long as there is no direct
support for this in the server. Otherwise, we can never make use of
this until we get servers supporting this feature. If we allow it,
then we can support xml:lang in clients *now* and get better support
for the future.
>The only "issue" with xml:lang is that in the examples in the XML
>recommendation they use "de-DE" instead of "de_DE".
>Other than that, this is precisely what xml:lang is meant for, and I
>see no reason not to make that the attribute we use.
>Reading you message again... if the xml:lang is done on the initial
>stream:stream, then the server can tack on xml:lang to the
><presence> and <message> in two cases:
>* if it's leaving this server, then the other server can determine
>if xml:lang is necessary
>* if it's staying on the same server, it can determine if the other
>user is in the same xml:lang or not
>This way, clients would see xml:lang only if it's *different* from
>their specified xml:lang.
Probably true, though not crucial right now. In fact, xml:lang will
probably ignored by most clients (except those which wish to hook in
a service like BabelFish for translation). Far more important is to
allow service components, like the JUD, to use the information to
send out localized forms.
>There are several ways to do this, I just want to see xml:lang
>specified with the initial stream connection (or maybe
>authentication), and would like it if the client didn't have to tack
>on xml:lang to everything, following the "let the server do the
Basically, I think this is a very good proposal. I am willing to help
with a JEP for this, and I'd instantly implement anything in this
area in JabberFoX (and Julian said the same for Gabber in the past).
email: <mailto:max at quendi.de>
phone: (+49) 6151-494890
More information about the Standards