[
https://issues.apache.org/jira/browse/LOG4J2-3372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17484742#comment-17484742
]
Gary D. Gregory edited comment on LOG4J2-3372 at 2/1/22, 4:23 PM:
------------------------------------------------------------------
It seems a console appender should specify its PatternLayout charset unless the
PatternLayout defines it explicitly.
This means our building framework needs access to a parent object when building
a child, in this case, the pattern layout would need to access the console
appender, but that is not how the configuration runs at all. Needs further
investigation. Otherwise, we need to document the Console Appender to ALWAYS
configure its layout with a charset.
Another concern ([~mattsicker] [~ralphgoers]) is that it should work as
expected, no extra config in master/3.x.
was (Author: garydgregory):
It seems a console appender should specify its PatternLayout charset unless the
PatternLayout defines it explicitly.
This means our building framework needs access to a parent object when building
a child, in this case, the pattern layout would need to access the console
appender, but that is not how the configuration runs at all. Needs further
investigation. Otherwise, we need to document the Console Appender to ALWAYS
configure its layout with a charset.
> ConsoleAppender on Windows does not use platform default encoding
> -----------------------------------------------------------------
>
> Key: LOG4J2-3372
> URL: https://issues.apache.org/jira/browse/LOG4J2-3372
> Project: Log4j 2
> Issue Type: Bug
> Components: Appenders
> Affects Versions: 2.17.1
> Reporter: Oliver Matz
> Priority: Minor
> Attachments: ConsoleEncodingExample.java, LOG4J2-3372-sceenshot.png,
> log4j2.xml
>
>
> The attached example outputs a short string containing non-ascii characters
> ("Schöner Spaß") both to System.out and via log4j2 to a ConsoleAppender with
> target SYSTEM_OUT.
> When I run the example in my IDE, it outputs the string correctly both times,
> as expected
> However, if I run the example in a console on my Windows Desktop (10.0), only
> the output to System.out is correct. The non-ascii-characters get broken in
> the output of log4j2.
> I know that a workaround is to add {{charset="IBM850"}} in the
> {{{}PatternLayout{}}}, but this clearly causes problems if the application
> runs in a different environment.
> I definitely expect that log4j uses the platform default encoding, just as
> System.out does, and this seems to be other peoples' expectation, too. See
> e.g. LOG4J2-2929, stating incorrectly that "Currently PatternLayout always
> uses the default charset if none is specified explicitly."
> Another workaround is, of course, to specify a certain encoding (e.g. UTF-8)
> both for System.out and for the PatternLayout and to require the user to set
> the windows codepage (using e.g. CHCP 65001). I dislike that approach because
> it defeats the purpose of the "platform default encoding" and places a burden
> on the user
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
