Commons Logging 1.1.0 ReleasePlan
This is an important release: JCL is very widely used in production but a combination of new specifications and widespread failures from vendors to adhere to the Java 2 classloading model have resulted in majors problems for the JCL 1.0.x series of releases.
Thanks to constructive criticism from Ceki Gülcü and new energy from new developers, changes have been made to the discovery model used by JCL. The current codebase is now more specification agnostic: it no longer insists on compliance with the classloader models introduced in Java 1.2 and in the servlet specification. It is believed by the team that the current codebase should solve those problems which can be solved given the design approach taken by JCL.
Preparations for Alpha 1
Most commons components use a release that uses multiple release candidates. Once the release is up to the technical standards required for Jakarta Commons releases, it is approved and released.
JCL 1.1.0 is an important release. Not only is JCL very widely used but the limitations of the 1.0.x series of releases has been widely reported. It is important that the 1.1 release not only resolves these problems but also does not introduce any new ones. It is also tricky to unit test situations involving exotic classloaders, so help from users of JCL is needed to verify that this new release works in unusual circumstances.
I'd therefore like to propose the following additions to the process. Release candidates will be produced as normal until one satisfies the standards required for commons releases. At that stage, though, the release will be dubbed commons-logging-1.1-alpha1. An announcement will be made requesting that users, developers and committers test this release. If there are no problems then a subsequent VOTE will be held to promote this release.
The release will be recut each time but no code modifications will be made without resetting the process.
Anyone who wants to volunteer for a task, just add something
Bug 31286 [logging] Memory leaks in JBoss due to LogFactory cache FIXED
Bug 32618 [logging] Enterprise Commons Logging : Globalization & more
- IBM's (through Richard) proposal which seems too much for this release.
Bug 35774 [logging] TCCL problem in J2EE Container
Bug 36041 [logging] Include class loader information when LogFactoryImpl throws LogConfigurationException. FIXED
- Reporter has been asked if it's OK to close this issue.
Bug 36062 [logging] extended API: getChildLogger(String)
- The two Joergs have said on the dev-list that they are willing to wait until a later release with this one.
Bug 36927 [logging] Disabling of TCCL Implemented, needs testing
Bug 37067 [logging] enhancement : add support for ant task logger
- Waiting for someone to create a patch.
Bug 37420 [logging] Online JCL 1.0.4 API Javadoc missing
This is a website issue. Someone needs to change a link. The process of copying the api-docs for a previous release should be documented, if it isn't already. FIXED
Bug 37427 [logging] Redirect stdout and stderr to logging system
Simon and Robert agrees that this should not go into commons logging. WONTFIX
Bug 37484 [logging] call to getClassLoader() in LogFactoryImpl not checked for null
- Might have been solved already.
Do we remove the ServletContextCleaner?
- It's obviously controversial. Maybe the code could be put in the documentation somewhere, or on the wiki.
Decide whether to merge the weak-hash-map stuff into the main trunk or leave it in an "optional" jar. If we merge it, we can do away with the optional jar completely which is good. However it does mean that if there is a bug in it people can't disable it. If bundled in the main jar there might need to be a little extra code to just ignore it when it throws an exception on load for java < 1.3.
- Sort out whether we split Log4JLogger into two classes or not. If we choose two classes, how should we name them?
Rename Log4J12Logger.java back to Log4JLogger.java. That would make the upgrade transparent for the previous use-case. But there is the chance that this will not work at all for a user that is currently using JCL 1.0.4 together with log4jalpha-something and a configuration file stating that Log4JLogger should be used.
- Users who configure JCL to use Log4JLogger might reasonably expect JCL to guess the log4j version and use the correct logger. so, perhaps one option would be to create a delegating implementation.
Same as #1 but kick out Log4J13Logger, because Log4J 1.3 has not been released.
- Decide our jar distribution strategy (in particular, whether we ship the optional jar or not).
- How do we give downstream packagers and users a fair view of the actual JCL dependencies?
Work To Resolve Design Issues
eliminate optional jar DONE
sub-components don't work very well. in particular, i think too many users are going to get too confused by yet another jar. WeakHashMap will go into the base distribution, other classes will be moved into contrib. perhaps another component (logging-extras) would be good or perhaps moving them off shore.
clean up source DONE
demonstration will be moved into contrib
- improve support for downstream packagers
add an ant task that creates a distribution with minimal dependencies. create guide to help people understand the distribution with section on dependencies.
- log4j loggers
log4j 1.3 is still not released. the new JCL release cannot depend on unreleased code. the 1.3 implementation will be moved into contrib. Log4JLogger and Log4J12Logger will be shipped with notes that direct use of log4jlogger is deprecated and will be replaced by a logical logger when log4j 1.3 ships.
this will be shipped. The critical issue was that projects which repackage JCL (eg by compiling with gcj) may not have an implementation of javax.servlet to compile against. However the commons-logging-api.jar project now contains all the stand-alone loggers, and this should be enough for most circumstances. Because such "repackagers" can create and distribute commons-logging-api.jar, it has been decided that it is acceptable for the commons-logging.jar to have a compile dependency on javax.servlet.
IoC friendly design postponed
Verify that TRACE support works for Log4J 1.2.12+. DONE
Process Bugs Marked Later
Up to Logging