In this case I am only translating a Java properties file, the text from
which gets pulled at runtime, but does not get compiled. I am not having
some of the difficulties with characters not displaying at all (as described
in a couple of bugparade reports) but I am having some difficulty with the
incorrect characters displaying as a result of using Native2ASCII. My
initial guess given that I have not had time to play with it is that I am
incorrectly specifying the source codepage, and thus the incorrect Unicode
numbers are being chosen.
-----Original Message-----
From: Adrian Havill [mailto:havill@threeweb.ad.jp]
Sent: Wednesday, May 06, 1998 12:52 AM
To: Unicode List
Subject: Re: Java 1.1.2 Native2ASCII
> Java 1.0.x is unsuitable for i18n work. Nominally, "char" was based on
> Unicode 1, but in reality, many of the classes had built-in Latin-1
> assumptions and will munge any Unicode chars that have the nerve to put
> something besides 0 in the high-order byte.
>
> For this reason, there was no Native2ASCII in any version of JDK 1.0.x,
and
> never will be, AFAIK.
Agreed. The compiler has errors in it as well in regards to handling string
literals, comments, and identifiers, so that if you use the "native2ascii"
on a
*.java file and then compile it with the 1.0.2 compiler, the compiler may
throw
UTF8Exceptions. (Even though there is no UTF-8 code in the source... I guess
it's because that was the closest thing to a "UnicodeException") In other
words,
certain valid Unicode character escapes (\u____) within constants or
characters
will spook the 1.0.2 compiler.
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:40 EDT