In [1], the fact that threads are executed unevenly is just a corollary of the "no assumption about the scheduler" principle. It does not mean that threads are starved. Looking at the way each task advances when LBCORE63 application is run, I could not detect signs of starvation. [1] http://jira.qos.ch/browse/LBCORE-96?focusedCommentId=11116#action_11116 Ceki Gulcu wrote:
Running LBCORE63 located in the ch.qos.logback.classic.issue package with JDK 1.6 seems to work nicely, at least so far according to tests performed on Windows and Linux. I can't see no starvation.
Joern Huxhorn wrote:
It's essentially the same situation as in my SynchronizedVs... app. It's just that the appender is the point of starvation in that case, not some Object.
-- Ceki Gülcü Logback: The reliable, generic, fast and flexible logging framework for Java. http://logback.qos.ch