Re: [xml-dev] xml:lang how often used?

["bryan rasmussen" <rasmussen.bryan@gmail.com>]

>
> It does seem an odd way of doing it - does anyone really have:
>
> <dictionary xml:lang="en">
>  <term name="Computer">Computer</term>
>  ...
> </dictionary>
> <dictionary xml:lang="fr">
>  <term name="Computer">Ordinateur</term>
>  ...
> </dictionary>
> <dictionary xml:lang="de">


I think it seems like a quick hacky way of doing it. Like, oh we
should have a dictionary of terms.
> ...where all of the various languages are included in r the one XML file?
Sure, Not well thought out dictionaries, but for just small
dictionaries of values for a small project I bet there's a lot.

> Wouldn't it be a better solution to do it the Java i18n way eg:
>
> <dictionary>
>   <term name="Computer"><i18n id="Computer"/></term>
>
> and then a language file for each locale:
>
> xml_en_UK:
>   <val id="Computer">Computer</val>
>
> xml_fr_FR:
>   <val id="Computer">Ordinateur</val>
>

I think that is not a preferable way because it is still using one
language as the default, the english one is redundantly maintained
across all files. Becomes also problematic if a word in some language
should be used that does not have a synonym in English. If the data
may become really big you need to do it on a network.

 How about like this:

> xml_en_UK:
>   <val id="http://mydictionary.org/5345";>Computer</val>
>
> xml_fr_FR:
>   <val id="http://mydictionary.org/5345";>Ordinateur</val>

Then a request against

http://mydictionary.org/5345?lang=en_uk
returns something like a wordml rdf format for Computer (I say
something like because I would probably want to do something else)

And http://mydictionary.org/languages/english/computer
should return  a redirect to
http://mydictionary.org/5345?lang=en_uk

and checking if a language was supported for a particular word
http://mydictionary.org/5345?lang=ar_sa
which might show that it was not supported or just do a redirect to
http://mydictionary.org/5345?lang=ar


Of course http replaced by whatever lookup scheme used in your system,
but I would think something like that more maintainable.

Cheers,
Bryan Rasmussen