Locale and strerror in libgcj's native code...

Tom Tromey tromey@redhat.com
Wed Aug 11 23:58:00 GMT 2004


David> It seems to me that using a mix (as we currently do) is clearly
David> wrong and only works in with ASCII encoding.

Yeah, I agree.

Bryce> Most POSIX OS's seem to be standardizing on UTF8, so I think its fine
Bryce> to assume UTF8 in these places. We can always change this later if it
Bryce> turns out to be a problem.

We also punt on this problem for file names.  There's a PR for this,
plus a lengthy thread or two in the java-patches archives.

I think it would be better if we had some new function (-like object)
for creating these strings; it could start life as a #define.
Certainly the current inconsistency is ugly, and if we ever wanted to
change things, it would be a pain.  A new name would both simplify
future changes and make hacking libgcj easier (the decision about what
function to call in these situations would be more clear).

I don't recall anybody ever complaining about this.  But I suspect
we've just been lucky.

Tom



More information about the Java mailing list