On Apr 17, 2005, at 1:43 PM, Ceki G�lc� wrote:
At 08:33 PM 4/15/2005, you wrote:I've been distracted from log4j for a while by log4cxx and more recently tax preparation. I hope to be able to spend devote some time to log4j in the near future. In my working copy I have a couple of new tests that I expect to commit before I start anything new:
EncodingTest was written primarily to support the implementation of WriterAppender.encoding in log4cxx. The code under test in log4j is fairly trivial, essentially the getters and setters, since encoding itself is handled by java.io.Writer. However, the code under test in log4cxx is fairly substantial and I really like having similar tests in log4j and log4cxx.
Why? log4cxx and log4j are distinct code bases. As far as I know, you are the only person who looks at both. If the implementation of a particular feature is done differently in log4j and log4cxx, then why should the test cases need to be similar?
Having nearly identical tests (when the structure of the code is similar enough to allow that) makes it much easier to make sure that log4j and log4cxx behave similarly under the same conditions. Sometimes the resulting log4j tests are testing behavior that you could take for granted (like java.io.Writer's encoding support) and would appear to be overkill for log4j's needs. However, the implementation may be much more shaky in log4cxx and having similar tests in both log4j and log4cxx makes it easier to ensure that they behave similarly.
LoggingEventTest primarily checks the serialization of LoggingEvent's against reference copies. I expect these tests to evolve as I get into the whole sequenceCount issue. However, at the present they do serve as a check against unintended changes in the serialized form.
I'd like to get these in before the switch to Subversion, so please let me know when that is imminent.
The migration to subversion should take place in the very near future. So it should be considered imminent.
Thanks.
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
