Re: multilingual tags

On 15/09/06, Warren Baird <photogeekmtl gmail com> wrote:
Dotan Cohen wrote:
> A workaround would be to assign to each tag a random identifier that
> is stored in the tag but not shown in the interface. The same
> identifier could be used for each language, and to store the language
> code, in the tag. So the actual tags would be "7484567::en::dog" and
> "7484567::sp::perro". When Fspot in in English mode, it would use the
> tag marked "en" and in Spanish mode would use the tag marked "sp".
> Another workaround would be to just have both languages in the same
> tag, like this: "en::dog::sp::perro".
> Both workarounds allow for more than two languages, should the user be
> so inclined. A solution would have to be drawn up as to what to do
> with tags that are labeled in only one language. I could provide ideas
> on that, if none are presented here.

The big concern I have with both of these options is interoperability
with other apps...   If I export my photos to flickr or picassa, or some
other photo app --- I would want the tags to reasonably sane at least...

One possibility that might be more reasonable would be to have dog,
perro, etc as separate tags the way we do today, but add some additional
information to the xmp data indicating that dog is an english tag, perro
is spanish, etc., and I guess it would also need to indicate
correlations, so if dog is deleted while english is the active language,
perro gets deleted as well.

I'm still not that familar with the xmp spec --- is there a way we can
embed info like that so that f-spot can find it, but it won't mess up
other apps?     That way if you view the file in flikr, you'd see both
dog and perro (which isn't too bad), buf if you view it in f-spot in
english, you see dog, in spanish you see perro, etc.

It's unclear to me how you'd produce a sane GUI for this, I must admit...


I like your idea of the XMP tags containing only the tag values so
that other apps will not be messed up. That means that the
correlations must be kept elsewhere- how about after the XMP data
there would be FSF data (F-spot Specific Format), that would contain
the correlation information. So the XMP data would contain, for
example, four tags, "dog", "perro", "one" and "uno". Then the FSF data
would look ike this:

       <value lang="en">Dog</value>
       <value lang="sp">Perro</value>
       <value lang="en">One</value>
       <value lang="sp">Uno</value>

In essence, the FSF data would suplement the XMP data. Other
applications, that do not recognise FSF data, would simply ignore it.
They would then just read the XMP data and find both Dog and Perro.

Dotan Cohen

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]