Insightful comments and suggestions. I'll thin my current environment a variable at a time and let you know when the noise level becomes low enough for the default build definition and process to work properly, with a goal of identifying the precise evironment variable(s) causing the problem, if possible.
I mentioned I recently rebuilt my system. I've collected my environment settings over several stable years and copied them faitfully to the new setup, but I don't know that all or any are required for the new environment to operate properly. Reducing the environment to the minimum required will be a good exercise. My Pivot 2.1 build works fine, by the way. I haven't tested every facet of it, but the testing I've done, including verification that the Expanders problem was fixed, has been successful. I use Oracle's Java distribution. On 9/17/12, Sandro Martini <[email protected]> wrote: > Hi Greg, > thanks for the info but it's really strange ... > I don't remember to have JRE_HOME set in any of my VM ... could be > this to generate some noise with other env variables. > > Did you tried with OpenJDK 7, or are you using Oracle JDK 7 ? > (this is another info to put in build output) > > I suggest you to make a custom javaEnv.sh script where to put all your > set/unset env vars, and run it before ecexuting any shell command. > In many VM I have this and it's useful ... for example see this: > http://svn.apache.org/repos/asf/pivot/infra/bin/setup-environment-sample.sh > > > All our CI Builds (on ASF Jenkins farm) are mostly on Linux servers > (ubuntu, and others) and without changes to build.xml all works, even > with Java 7 . Just for reference: > https://builds.apache.org/job/Pivot-trunk/ > https://builds.apache.org/job/Pivot-trunk%20on%20Java%207/ > > And when I make a Pivot release, I use an ubuntu VM with a standard > installation of JDK, but with JDK 6. > > Anyway thanks for the info now, I'll try to take a look at it with a > Linux VM ... > > > Bye, > Sandro >
