This is the mail archive of the
mailing list for the Java project.
Re: Freenet compilation errors.
>>>>> "Oskar" == Oskar Liljeblad <firstname.lastname@example.org> writes:
Oskar> This might've been brought up before, but shouldn't this be the
Oskar> default? At least it seems to be for Sun javac.
When I implemented this I chose to make the default come from the
user's locale. Maybe there is a way to override this when the user
has the default (C) locale. I'm not sure.
On Linux boxes the C locale uses ASCII and iconv() will return errors
for characters > 127.
In other words, we really just depend on the system to tell us what to
Oskar> Or is there any environment variable I can set to make this the
Oskar> default? (Wouldn't GCJ be able to import my LC_CTYPE=en_GB,
Oskar> which allows ls to print 8859_1 characters?)
Yes, this ought to work right now.
For instance on my box I can use "LANG=en_US.UTF-8 gcj ..." to tell
gcj that I want the UTF-8 encoding by default. (LC_CTYPE works as
well.) Someday I really should investigate running UTF-8 all the
time. I hear it is possible, or nearly so, these days.