Locale sets

[Kelly O'Hair]

As I recall.  There were different rules in place as to what messages had to be localized, based on being in the JRE, the JDK, or the plugins
which were unique in terms of being run from different browsers. I don't know what the rules are now.
I think these resource files represent 'expected' translations that would need to be made, but it's been a while since I looked at this logic.
It is messy.

The JRE and plugin didn't have near as many messages as the JDK.
Usually the JRE and the plugins just operated at runtime with few messages seen.

The JDK messages of primary interest were those from the javac compiler and command line utilities, which for the most part
were developer tools. But there were many messages in the JDK, maybe 10X or 100X what is in the JRE as I recall.

There is a significant cost to doing the translations of these messages, especially compiler type messages where the
phrasing and context is so messy in terms of localization. 

Not sure this helps much...

-kto

On Jan 4, 2012, at 5:31 AM, Erik Joelsson wrote:

> Hello,
> 
> I have a question about the old (current) build system. There are resource files for various locales, most of them are present explicitly in the source while others are generated based on another. The makefiles are aware of the set of locales being included in the build. The set can be limited by either providing a complete set yourself in a make variable or by choosing between three predefined sets called "jdk", "jre" and "plugin". I'm wondering what the reason is for wanting to limit the amount of locales being built and if anyone is using this feature. Also, what's the purpose of the three predefined sets?
> 
> /Erik