[jira] [Commented] (LOG4J2-1873) Implement UTF-8 encoding that doesn't use CharsetEncoder
[ https://issues.apache.org/jira/browse/LOG4J2-1873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15963846#comment-15963846 ] Roman Leventov commented on LOG4J2-1873: Or implementation from Apache Harmony: https://github.com/apache/harmony-classlib/blob/15867416ca717a9910abb787989205223f5b1f09/modules/nio_char/src/main/java/org/apache/harmony/niochar/charset/UTF_8.java#L219 > Implement UTF-8 encoding that doesn't use CharsetEncoder > > > Key: LOG4J2-1873 > URL: https://issues.apache.org/jira/browse/LOG4J2-1873 > Project: Log4j 2 > Issue Type: Improvement >Reporter: Roman Leventov > > CharsetEncoder accepts only CharBuffers, and for the sake of being entirely > garbage-free we don't want to use CharBuffer.wrap(stringBuilder), when > encoding an event. > That forces us to make additional data copy from StringBuilder to a > thread-local CharBuffer. > This could be avoided by implementing UTF-8 encoding logic in log4j-core > itself, and not using CharsetEncoder. > This issue is specifically about UTF-8 because it is used predominantly and > it's relatively easy to implement. There are also likely some open-source > Apache 2-compatible implementations in Java out there already that we could > just copy and adapt. -- This message was sent by Atlassian JIRA (v6.3.15#6346) - To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org For additional commands, e-mail: log4j-dev-h...@logging.apache.org
[jira] [Commented] (LOG4J2-1873) Implement UTF-8 encoding that doesn't use CharsetEncoder
[ https://issues.apache.org/jira/browse/LOG4J2-1873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15963050#comment-15963050 ] Gary Gregory commented on LOG4J2-1873: -- ICU4J might have something we can optionally depend on or copy. > Implement UTF-8 encoding that doesn't use CharsetEncoder > > > Key: LOG4J2-1873 > URL: https://issues.apache.org/jira/browse/LOG4J2-1873 > Project: Log4j 2 > Issue Type: Improvement >Reporter: Roman Leventov > > CharsetEncoder accepts only CharBuffers, and for the sake of being entirely > garbage-free we don't want to use CharBuffer.wrap(stringBuilder), when > encoding an event. > That forces us to make additional data copy from StringBuilder to a > thread-local CharBuffer. > This could be avoided by implementing UTF-8 encoding logic in log4j-core > itself, and not using CharsetEncoder. > This issue is specifically about UTF-8 because it is used predominantly and > it's relatively easy to implement. There are also likely some open-source > Apache 2-compatible implementations in Java out there already that we could > just copy and adapt. -- This message was sent by Atlassian JIRA (v6.3.15#6346) - To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org For additional commands, e-mail: log4j-dev-h...@logging.apache.org
[jira] [Commented] (LOG4J2-1873) Implement UTF-8 encoding that doesn't use CharsetEncoder
[ https://issues.apache.org/jira/browse/LOG4J2-1873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15962930#comment-15962930 ] Matt Sicker commented on LOG4J2-1873: - I'd be interested in seeing this feature myself. The standard UTF-8 CharsetEncoder is ridiculous in this regard. > Implement UTF-8 encoding that doesn't use CharsetEncoder > > > Key: LOG4J2-1873 > URL: https://issues.apache.org/jira/browse/LOG4J2-1873 > Project: Log4j 2 > Issue Type: Improvement >Reporter: Roman Leventov > > CharsetEncoder accepts only CharBuffers, and for the sake of being entirely > garbage-free we don't want to use CharBuffer.wrap(stringBuilder), when > encoding an event. > That forces us to make additional data copy from StringBuilder to a > thread-local CharBuffer. > This could be avoided by implementing UTF-8 encoding logic in log4j-core > itself, and not using CharsetEncoder. > This issue is specifically about UTF-8 because it is used predominantly and > it's relatively easy to implement. There are also likely some open-source > Apache 2-compatible implementations in Java out there already that we could > just copy and adapt. -- This message was sent by Atlassian JIRA (v6.3.15#6346) - To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org For additional commands, e-mail: log4j-dev-h...@logging.apache.org
[jira] [Commented] (LOG4J2-1873) Implement UTF-8 encoding that doesn't use CharsetEncoder
[ https://issues.apache.org/jira/browse/LOG4J2-1873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15962650#comment-15962650 ] Roman Leventov commented on LOG4J2-1873: [~rem...@yahoo.com] Yes, I think it would be something like "Utf8StringBuilderEncoder". > Implement UTF-8 encoding that doesn't use CharsetEncoder > > > Key: LOG4J2-1873 > URL: https://issues.apache.org/jira/browse/LOG4J2-1873 > Project: Log4j 2 > Issue Type: Improvement >Reporter: Roman Leventov > > CharsetEncoder accepts only CharBuffers, and for the sake of being entirely > garbage-free we don't want to use CharBuffer.wrap(stringBuilder), when > encoding an event. > That forces us to make additional data copy from StringBuilder to a > thread-local CharBuffer. > This could be avoided by implementing UTF-8 encoding logic in log4j-core > itself, and not using CharsetEncoder. > This issue is specifically about UTF-8 because it is used predominantly and > it's relatively easy to implement. There are also likely some open-source > Apache 2-compatible implementations in Java out there already that we could > just copy and adapt. -- This message was sent by Atlassian JIRA (v6.3.15#6346) - To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org For additional commands, e-mail: log4j-dev-h...@logging.apache.org
[jira] [Commented] (LOG4J2-1873) Implement UTF-8 encoding that doesn't use CharsetEncoder
[ https://issues.apache.org/jira/browse/LOG4J2-1873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15962607#comment-15962607 ] Remko Popma commented on LOG4J2-1873: - I'm open to ideas. We did something similar with core.util.StringEncoder for 8859-1 but that's not used on the garbage-free path. Would this be an enhancement or replacement of StringBuilderEncoder or something else entirely? > Implement UTF-8 encoding that doesn't use CharsetEncoder > > > Key: LOG4J2-1873 > URL: https://issues.apache.org/jira/browse/LOG4J2-1873 > Project: Log4j 2 > Issue Type: Improvement >Reporter: Roman Leventov > > CharsetEncoder accepts only CharBuffers, and for the sake of being entirely > garbage-free we don't want to use CharBuffer.wrap(stringBuilder), when > encoding an event. > That forces us to make additional data copy from StringBuilder to a > thread-local CharBuffer. > This could be avoided by implementing UTF-8 encoding logic in log4j-core > itself, and not using CharsetEncoder. > This issue is specifically about UTF-8 because it is used predominantly and > it's relatively easy to implement. There are also likely some open-source > Apache 2-compatible implementations in Java out there already that we could > just copy and adapt. -- This message was sent by Atlassian JIRA (v6.3.15#6346) - To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org For additional commands, e-mail: log4j-dev-h...@logging.apache.org
