localization

Dmitri Dolguikh dmitri at redhat.com
Thu Apr 15 17:08:28 UTC 2010


On 10-04-15 2:01 PM, Bryan Kearney wrote:
> Cool.. couple of questions:
>
> On 04/15/2010 12:38 PM, Dmitri Dolguikh wrote:
>    
>> changes are in. Here's the skinny:
>>
>> - there's a new dependency in the build: gettext-commons
>> - there are new targets in the build: extractkeys - extracts strings
>> that need to be localized, mergekeys - performs a merge on the existing
>> .po files, createresourcebundles - compiles resource bundles
>> - localized resource bundles can be found in
>> localized_resources/classes/ directory
>>      
> Is there a way to identify the keys which have not yet been translated?
>    
look at the actual .po file and see what msgstr's are empty (and what 
keys they correspond to). I can do a buildr task for that.
>    
>> - if you're running tomcat from under eclipse, you'll need to the above
>> directory into tomcat's classpath
>>
>> how to use it (development):
>> inject I18n (there's a I18n provider) instance in the classes that will
>> have user-visible text. Then (assuming instance variable name is i18n):
>>      
> I assume still open to do is to yank the locale off of the current request?
>    
Nope. done. (see I18nProvider)
>    
>> <snip>
>> i18n.tr("goodbye, cruel world.");
>> </snip>
>>      
>
> So. best practice is to use tr("string with {0}", variable) ??
>    
correct.
> Also... I assume since it is scanning for .tr() we can not put the i18n
> into the actual exception class?
>    
+1. that's what i discovered earlier this week.
> I doubt we could do BadRequestExcepotion.tr("Some text"). Seems like a
> hack :)
>    
we could. but i think the current way of dealing with it is cleaner 
(keep in mind, we'll have to get the locale from the request in any 
case, and that gets messy).

-d



More information about the candlepin mailing list