I have noticed that, in contemporary languages with easy dynamic allocation and garbage collection, we wind up with a lot of "caches" that aren't caches in the traditional sense, but baskets that fill slowly and endlessly with everything we've ever seen. This approach (e.g. allocate a HashMap and squirrel away every expensive object we have to generate) works well in a narrow scope where the whole cache will soon be discarded, but I feel that global caches like this are trouble waiting to happen. A global cache needs some mechanism which *regularly* evicts entries.
I can't recall ever coming across a stock Java class that implements limited-count caching with LRU eviction, for example. I think anything we can do to reasonably limit the lifetime of cache entries will be an improvement that costs little in either performance or complexity. -- Mark H. Wood, Lead System Programmer [email protected] Asking whether markets are efficient is like asking whether people are smart.
pgpxJHwF8cSGV.pgp
Description: PGP signature
------------------------------------------------------------------------------ All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity and more. Splunk takes this data and makes sense of it. Business sense. IT sense. Common sense. http://p.sf.net/sfu/splunk-d2d-oct
_______________________________________________ Dspace-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dspace-devel
