Gabriel Roldan wrote:
> Arne Kepp wrote:
>> Sorry, reading the email again, I realize I got caught up in the RBE 
>> / Eclipse part of the discussion:
>>
>> When I started this, at least 30% of the strings in the application 
>> were hardcoded in English. I made complete templates with the (DE) 
>> prefix in order to spot all the places where the strings were not 
>> read from resources at all (-> they would be missing the (DE) prefix 
>> in the UI).
> sure, that makes sense.
>>
>> The complete templates and (DE) prefixing was not meant as an example 
>> for future i18n work. So the answer to your original question is no, 
>> I did not intend that you would have to add strings to all the 
>> translations. That said, we will need at least one "complete" 
>> translation, to flush out when people mess up and commit hardcoded 
>> strings.
> ok, we can take the spanish one as the "complete" one. I already 
> spotted some hardcoded strings that still need to be i18n'ized and am 
> meaning to keep it in sync.
>>
>> The other important thing I did was to list all the strings in the 
>> default translation in an orderly manner. So that if the default file 
>> is expanded, one only needs to look at the default and the 
>> translation side by side to spot where strings are missing, without 
>> installing Eclipse.
> point taken. Maybe we can come up with a bash script that reports what 
> the missing resources are?
>>
>> This is the ability that I'm afraid we will lose with tools like RBE. 
>> Do we?
> Yes, we would. It doesn't seem to be a perfect approach. Another 
> concern wrt i18n is no longer used resources. Ideally we should also 
> keep track of non used resources so the bundle is not incrementally 
> filled with them. That and modifications that might imply the 
> translation needs to be modified too. Hopefully it's not that a big 
> concern and tending to settle down as the UI stabilizes.
>
> So if I have to conclude, I would say:
> - let each translator choose the procedure he feels better about, as 
> long as: the default locale gets not mangled, so that's the one for 
> which svn history matters.
> - A script to spot out the missing translations would be helpful.
> - When adding a new resource to the default locale do not add it to a 
> translation at least you're actually translating
> - when no longer using a resource take the time to remove it from the 
> default and all the translations?
> - when modifying a resource key modify it also on the translations
> - when modifying a resource value delete it from the translations?
>
> That's just a proposal, may be too bothering. What do you think?
>
> Cheers,
> Gabriel

Sounds good, especially the last three rules. Maybe we can steal a 
language "nobody will ever use (tm)" and use it to test for hardcoded 
strings, i.e. automatically copy the default and prefix with something.

Maybe the script can be a community module that we run with maven ? 
Quickly looked for examples using awk, looks like it'll be easier to do 
this in Java than in bash.

-Arne

-- 
Arne Kepp
OpenGeo - http://opengeo.org
Expert service straight from the developers


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Geoserver-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

Reply via email to