On 9/5/06, Paulex Yang wrote:
Stepan Mishura wrote: > Paulex, thanks for your patience and answers. > > Yes, Sun's JRE sets root logger's level according to > 'lib/logging.properties'. Also specifying the level for root logger in > j.u.l.config.file doesn't help. (I assumed before that a JRE respects my > "lovely" custom LogManager settings :-(! ) > > OK. Let's return to the initial issue: how to run logging tests without > changing JRE config files? > > I still believe that there should be some elegant way for testing logging > implementation without touching JRE files. I agreed this is not acceptable. AFAIK current LogManagerTest has used a MockLogManager in most test cases, so the migration should be relatively easy. > For example, what > about developing custom LogManager for testing? The testing manager will > work in 2 modes: default configuration and custom configuration. A > test will > have a hook to switch testing manager's mode. The first configuration is > read from JRE config files (by j.u.l.LogManager implementation) and the > second one will be fully controlled by a test and is set by the test > via some manager's method (say readConfiguration(InputStream)). Does this > sound crazy? Maybe we just need one Support_Exec() invocation to verify if the LogManager reads the default logging properties(jre/lib/logging.properties) correctly(which should be straightforward by also loading and parsing that properties in test codes), for other tests, we always use customized LogManager or configuration files. If you are fine with this, I'm volunteer to refactor the tests. Agree with this approach.
Thanks, Stepan ------------------------------------------------------ Terms of use : http://incubator.apache.org/harmony/mailing.html To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
