OpenEuphoria

Re: international language support

• Posted by Irv Mullins <irvm at ellijay.com> Aug 16, 2001
• 462 views

On Thursday 16 August 2001 11:14, freeplay at mailandnews.com wrote:

> Parsing isn't the issue with NLS.  NLS is a way to help you code an
> application once (single source) and have it run in as many different
> languages you have the time to translate a message catalogue into.  Not
> sure if it handles reading text right to left.  I guess you just put the
> message text in the catalog right to left.  The other biggie is character
> sets.

Parsing would be an issue if keywords were translatable.
If you look at any number of Linux programs which can run in multiple 
languages, you'll see that the source code still uses "English-ish" 
keywords. Lots of if ()  printf(), etc.

I think Logo has been translated into French. and Applescripts worked 
when written in French.

> I've never coded anything with NLS so I'm no expert.  I'd guess, however,
> that the format of a message catalogue has an "english" feel to it though

Pretty much  the same thing could be done in Euphoria with simple include 
files:
msg.en
msg1 = "File not found"
msg2 = "Cannot read file"

msg.es
msg1 = "Fichero no encontrado"
msg2 = "No puede leer el fichero"

(Apologies for the poor spanish)
Just include the desired msg.xx file for your language.
The advantage would be that anyone, not just a programmer, who can 
read one of the existing msg. files could easily translate that into a new 
msg. file in her native language, and your program would be usable.

One disadvantage might be extra work for the programmer, having 
to keep track of all the msg id's, and always using them instead of just 
typing the text in directly.  A pre-processor might be able to 
solve this, by scanning the finished code, extracting and creating a list of 
all the unique strings, saving that list to a msg. file, and then replacing 
those strings in the source with the msg1 .... msg2... etc. pointers.

Of course, this doesn't help much with the other problems, such as 
the translated string being much longer than the original, 
English => German, for instance. 

Regards,
Irv