Hi Yann, On Fri, Jun 23, 2006 at 12:49:58AM +0200, Yann Dirson wrote: > The 1.3 NMU has a problem, as shown by the results of jvm-test in the > log. Because some time ago the "mmap" test was known to fail, the jvm > tests are not fatal to the build. Now sablevm/classpath have > improved, ant we can make this test fatal again. This should make > -1.3 fail on every jvm-enabled arch.
Of course, it's your prerogative to not want this version of bigloo to ship with etch. However, the reason that the JVM tests fail at build time in 1.3 is that there is no appropriate JVM *installed*: > And in -1.3: > Recette (JVM) Done... > ------------------------------- > make[3]: Leaving directory > `/build/buildd/bigloo-2.7a/build-tree/bigloo2.7a/recette' > Exception in thread "main" java.lang.ClassFormatError: main (erroneous class > name) > at java.lang.VMClassLoader.defineClass(libgcj.so.7) > at java.lang.ClassLoader.defineClass(libgcj.so.7) > at java.security.SecureClassLoader.defineClass(libgcj.so.7) > at java.net.URLClassLoader.findClass(libgcj.so.7) > at java.lang.ClassLoader.loadClass(libgcj.so.7) > at java.lang.ClassLoader.loadClass(libgcj.so.7) > at java.lang.Class.forName(libgcj.so.7) > at gnu.java.lang.MainThread.run(libgcj.so.7) > make[2]: *** [jvm-test] Error 1 I verified on my system, before proposing this patch, that the test suite passes if I re-run it after installing sablevm; i.e., there is no problem with the resulting bigloo binaries, they just are not compatible with the gcj runtime due to problems with gcj itself. So this seems to me to be a very straightforward workaround for the problem of sablevm being in an inconsistent state across architectures right now in unstable, even though the packages in testing should work fine. Another possibility though would be to make bigloo-backend-jvm arch: all, since it only contains java code. That should save us around 15MB in the archive (per version), and it means not having to worry about which architectures sablevm is working on at any given moment when building. Would you find that option more acceptable? Thanks, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/
signature.asc
Description: Digital signature
