Re: JavaMemoryModel: A different slant with reference leakage

From: Sylvia Else (sylviae@optushome.com.au)
Date: Sun Jan 18 2004 - 18:21:38 EST


I should probably add that by "correct" I mean correct within the context
of the particular memory model in use. Thus the model has to be such that
the developer can reasonably easily understand why the first solution isn't
correct, and can with similar facility deduce that the second solution is.

Note that there is nothing fancy about this code. The potential leak does
not arise from causality violations. Yet it feel that this example is much
more typical of a real life scenario.

Sylvia.

-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel



This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:56 EDT