I think we have been making our first impression - we are afraid to ever say it is good enough and always want the freedom to break compatibility, so don't use our code.
Sent from my iPad
On Jul 15, 2014, at 4:27 AM, Gary Gregory <firstname.lastname@example.org> wrote:On Tue, Jul 15, 2014 at 6:47 AM, Remko Popma <email@example.com> wrote:
About the proposed change for LOG4J2-703/713: I don't see any reason why this fix can't go in a 2.0.1 or 2.1 release.In general, I think we agree that in any release log4j-core can have changes that are not binary compatible with previous releases. That is why the api and core modules are separate: so we are free to make changes to core...To be honest, LOG4J2-703/713 doesn't look like a showstopper, or am I missing something?My concerns are twofold:- The Closer API has changed. The is a 'public' API and breaks BC; whether or not we consider this class as public for the purpose of defining our semantic versifying is a topic we need to address and document to set expectations for our users. If we say 'classes in package foo' should not be used by non-log4j-modules, then we are drawn a line in the sand. The fact that we do not put these classes in an 'internal' package a la Eclipse is a different topic also.
- First impressions. It would be nice to address easy bugs before 2.0 goes out; this seems to be an easy bug. Yes, it could be in 2.0.1 or 2.1 but you only make a first impression once.Gary
Sent from my iPhone
On 2014/07/15, at 19:20, Gary Gregory <firstname.lastname@example.org> wrote:
Note that while 703 is marked as resolved, the user is still having the problem, so I will re-open it, and I would say another RC is needed to remove 703 from the generated JIRA report/release notes.
In the meantime, I will attempt another round-trip of fix/test with the user.GaryOn Mon, Jul 14, 2014 at 11:34 PM, Gary Gregory <email@example.com> wrote:
Since 703 is resolved, I created https://issues.apache.org/jira/browse/LOG4J2-713 and committed a fix to trunk.
I wonder what else will pop up on Android. It looks like one of our JDBC classes also depends on JNDI so that would bomb too.
Perhaps we should delay voting on 2.0 until we know how 703 and 713 play out. Especially since splitting classes in two might be the only solution.
GaryOn Mon, Jul 14, 2014 at 11:20 PM, Gary Gregory <firstname.lastname@example.org> wrote:GarySo the question is: Are we, and if yes, what modules, allowing ourselves to break BC in a non-major release.I have a simple fix for https://issues.apache.org/jira/browse/LOG4J2-703 "Could not find class 'javax.naming.InitialContext', referenced from method org.apache.logging.log4j.core.lookup.JndiLookup.lookup".This breaks BC in org.apache.logging.log4j.core.util.Closer.
On Sat, Jul 12, 2014 at 8:25 PM, Ralph Goers <email@example.com> wrote:
This is a vote to release Log4j 2.0, the first GA release of Log4j 2.Please test and cast your votes. +1, release the artifacts -1, don't release because…The vote will remain open for 72 hours (or more if required).New features:o LOG4J2-519: Added support for generating custom logger wrappers that replace the existing log levelsand extended logger wrappers that add custom log levels to the existing ones.o LOG4J2-696: RegexFilter does not match multiline log messages.Fixed Bugs:o LOG4J2-705: Fixed issue where Async Logger does not log thread context stack data.API change: added method getImmutableStackOrNull() to ThreadContext.ContextStack interface.o LOG4J2-631: Update docs to clarify how to use formatter logger and standard logger together.o LOG4J2-441: LoggerConfigs with no Level now inherit the Level from their parent.o LOG4J2-703: Android: Could not find class 'javax.naming.InitialContext', referenced from method org.apache.logging.log4j.core.lookup.JndiLookup.lookup. Thanks to Nelson Melina.o LOG4J2-699: PatternLayout manual page missing documentation on header/footer.o LOG4J2-625: Fixed Serialization error with SocketAppender and Async Loggers.(Fixed in RC2, but wasn't included in release notes.)o LOG4J2-538: JMX GUI: fixed occasional ArrayIndexOutOfBoundsException after pressing "reconfigure with XML below".(Fixed in RC2, but wasn't included in release notes.)o LOG4J2-666: AsyncLoggerContextSelector should ensure that different AsyncLoggerContext objects created by web app classloaders have unique names.o LOG4J2-683: Fix annotation processor warnings on JDK 1.7+. Thanks to Jurriaan Mous.o LOG4J2-694: Fix strange compilation error that popped up in a test class.o LOG4J2-692: Update documentation to specify only Maven 3 is supported.o LOG4J2-690: Log4j Web test dependencies should be in scope "test" in the pom. Thanks to Philip Helger.o LOG4J2-682: Special characters (tab and so on) in PatternLayout do not work. Thanks to Scott Harrington.o LOG4J2-686: Core's OptionConverter support for \b is broken (affects PatternLayout).o LOG4J2-687: Rename org.apache.logging.log4j.core.util.Closer.closeSilent() to closeSilently().o LOG4J2-688: Make org.apache.logging.log4j.core.layout.PatternLayout immutable.o LOG4J2-707: Some exceptions are not logged when configuration problems are detected.Changes:o LOG4J2-685: Make org.apache.logging.log4j.core.layout.AbstractLayout immutable.o LOG4J2-689: Update Jackson to 2.4.1.o LOG4J2-709: Update Apache Commons Logging to 1.2 from 1.1.3.SVN revision: 1610084Web Site: http://people.apache.org/~rgoers/log4j2/You may download all the artifacts by doing:wget -e robots=off --cut-dirs=3 -r -p -np --no-check-certificate https://repository.apache.org/content/repositories/orgapachelogging-1004/org/apache/logging/log4j/Nexus did not send an email. The list of artifacts can be found at the link above.