Hi Diego,

> Against my feeling I'll take in my hands the work to try the famous
> "Alternative B" (It means to make the translation in the widgets).

But why would you want to do this??? I completely agree with your original
POV - it's the model that needs to be translated not the view! Having "a few
general translation methods" in a few cases just isn't going to cut it -
there are way too many possible combinations of individual strings and the
"backend translation" will only lead to endless variations of composed
strings which absolutely require changes in the code to generally come out
correctly.

If you're only concerned about the "lines of code to touch" in the
translation process it may be worthwhile to consider merging our two
approaches - use mine for simple literal strings and don't touch code
explicitly and use yours in those cases where the translations are too
complex.

It may be interesting to see the percentage in which the simple literal
translation actually does work - can you check your changes to get a feel
for this?

Cheers,
  - Andreas

PS. Here's a snippet from the IRC party yesterday ("bander" = me; "gokr2" =
Göran)

14:14:33 <gokr2> bander: What are your thoughts on Babel?
14:16:01 <bander> gokr: Diego has it all nicely worked out - in fact it
models precisely the approach Mike and I were talking about.
14:16:45 <gokr2> bander: Great.
14:17:11 <gokr2> bander: I am asking since I know you did an approach a
while back that looked slick.
14:20:50 <bander> gokr: The approaches are identical in what they are trying
to achieve. The only difference is whether you have implicit or explicit
conversions. I don't care too much which way it goes - with implicit
conversions you are less intrusive in changing code but with explicit ones
you are being reminded that these things _will_ be user visible and require
some attention.
14:30:45 <bander> gokr: BTW, the approaches are not at all mutually
exlusive. You can see #translated as a message indicating that a string
should be translated. This information is useful even if String>>translated
is a no-op; you could analyze an image for all sends of translated and
simply collect the strings for the implicit translation. So the implicit
scheme can be used either with or without the explicit one and the "marker"
added by #translated is useful [for either one]


> -----Original Message-----
> From: squeak-dev-bounces at lists.squeakfoundation.org 
> [mailto:squeak-dev-bounces at lists.squeakfoundation.org] On 
> Behalf Of diegogomezdeck at consultar.com
> Sent: Wednesday, October 08, 2003 10:42 PM
> To: squeak-dev at lists.squeakfoundation.org
> Subject: [Babel] About alternative B?
> 
> 
> Hi Folks...
> 
> Against my feeling I'll take in my hands the work to try the famous
> "Alternative B" (It means to make the translation in the widgets).
> 
> Before going with the huge task I'd like to know what you're expecting
> exactly.
> We have a draft in the Scott Wallace's email (subject: Re: 
> [Babel] Looking
> for more translators).
> 
> What more is expected?
> 
> Cheers,
> 
> Diego Gomez Deck
> http://www.small-land.org
> 
> 
> 
>