Have you looked into the possibility that running your coverage tests might be run with or without instrumented code? (I don't know precisely how cobertura works: namely by instrumenting code at load-time or at compile-time, so please bear that in mind before taking my word in account :)).
Or something that could be close to the behaviour given by : surefire-report:report-only instead of surefire-report:report? My 2 cents. 2009/1/6 Harper, Brad <[email protected]> > No one replied to my initial posting, so on the chance that this issue > was overlooked by the *one* person who might be able shed some light on > it ... > > After exploring the possibility, I'm no longer inclined to think that > the presence/absence of debug info is involved. > > Brad > > > -----Original Message----- > > From: Harper, Brad > > Sent: Friday, December 19, 2008 4:20 PM > > To: 'Maven Users List' > > Subject: Inconsistent Cobertura Results > > > > Sometimes cobertura reports zero percent coverage in info > > produced by site:site, and non-zero percentages are reported > > when run separately. > > > > I was thinking that this might be due to differences in the > > default values of the property "maven.compiler.debug" between > > the build and site lifecycles ... or in the ways the compile > > plugin might be run for compilation of tests across these uses. > > > > Has anyone seen this behavior? > > > > BTW, I've seen "-Dmaven.compiler.debug=true" in postings and > > "${maven.compiler.debug}" in the plugin docs ... and also > > "-Dmaven.compile.debug=on" in postings. > > > > I'm assuming that the former is correct, and that the latter > > refers to Maven 1. > > > > Brad > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > -- Baptiste <Batmat> MATHUS - http://batmat.net Sauvez un arbre, Mangez un castor !
