Are FastHashMap, etc. safe considering JSR 133?

[David Morris <Da...@plumcreek.com>]

While discussing the protections afforded by various synchronization 
schemes, it was pointed out to me that on weakly consistent hardware  
what used to work may fail. I have used synchronization like the 
Jakarta Commons FastHashMap extensively and am now contemplating 
what this means. The developers of the hardware platform we target 
(IBM's iSeries) tell me they are moving to a weakly consistent memory 
model. Is the fix to avoid techniques like FastHashMap and simply 
synchronize all access in a multithreaded environment? How pervasive 
is the use of techniques like the FastHashMap in the various Jakarta 
projects like Tomcat, etc., which we use extensively?

Here is a reference that helps to explain the ramifications of JSR
133:

http://www.cs.umd.edu/~pugh/java/memoryModel/

Thanks,

David Morris

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>