On Jun 20, 2007, at 10:37 PM, Paul Smith wrote:
On 21/06/2007, at 1:27 PM, Curt Arnold wrote:
I've just pushed another release candidate out. This won't be the
last since I forgot to update the release date and another tweak
to the POM that I'd like to do. I'll likely refresh it tomorrow,
but I thought I would leave it up overnight for any immediate
feedback. The distributions are at http://people.apache.org/
builds/logging/log4j/1.2.15 and Maven artifacts are in the staging
repo at http://people.apache.org/build/logging/repo.
the repo url doesn't work for me.
Sorry, should have been "builds", not "build".
mkdir test
cd test
tar -xvzf ~/log4j/target/log4j-1.2.15.tar.gz
cd apache-log4j-1.2.15
jar xf log4j-1.2.15.jar
hexdump NTEventLogAppender.dll > NTEventLogAppender.hex
cd ../..
mkdir reference
cd reference
tar -xvzf ~/downloads/apache-log4j-1.2.15_rc3.tar.gz
cd apache-log4j-1.2.15
jar xf log4j-1.2.15.jar
hexdump NTEventLogAppender.dll > NTEventLogAppender.hex
cd ../..
diff -r reference test
Would running a jdiff report between 1.2.14 and 15 help?
Yes, but that is a different type of test. The idea here was that
one log4j release should be reproducible by different individuals in
case there is a need to a hotfix release and the previous release
manager was run over by that bus that keeps targeting developers.
Can't eliminate the different timestamps in the files, but the .class
files should be reproducible. When Mark Womack was release manager,
I had a duplicate build environment that I would use to replicate his
build and then do this type of test to make sure that I could come up
with an almost identical release if the need ever arose.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
