Arieh,
From: Arieh Markel <[EMAIL PROTECTED]>
Subject: Question - Re: cvs commit: DefaultCMSetter.java
Date: Fri, 16 Mar 2001 17:34:30 -0700 (MST)
Message-ID: <[EMAIL PROTECTED]>
> Shouldn't the charset be according to the Locale and Charset of the
> request that was passed ?
I think there are three localization (or internationalization) model
on servlet container:
1, Based on Java (= platform) default locale. This model is used for
internal modules which output native messages (for example, errors of
native libraries or applications etc.) or administration messages.
2, Based on the specified locale by service providers. This model is
general.
3, Based on the specified locale by users. This model is used for
servlets which use content nagotiation and output messages according
to their language requests.
The reason that we selected model "1" for DefaultCMSetter.java is to
enable error outputs. On japanese Solaris or Windows, tomcat sometimes
returns error messages in japanese but they aren't display correctly
in the case of content type is "text/html" (= "text/html;
charset=iso-8859-1"). We must change charset.
And another problem is naming scheme about Java character encoding
converters. Most of Java VM uses non MIME prefered name. Someone
thinks that it is better to use "file.encoding" property but it is
"eucJP" (cf, MIME prefered name is "EUC-JP") on Japanese Solaris 2.
We need normalization to MIME prefered names.
But I think there may be another solution. I hope you will do better
jobs.
Kazuhiro Kazama ([EMAIL PROTECTED]) NTT Network Innovation Laboratories
