Community
Participate
Working Groups
There are the following references to new ViewerSorter() in Help. This method creates a Collator using Collator.getInstance() which can take around 200ms (on my hyperthreaded machine). If you cache a collator you might get a good performance saving. ContextHelpSorter FederatedSearchSorter SorterByScore
Tod, the implementation of Collator.getInstance() already uses cache. Why would my own caching improve performance?
Ah, I think I know - they do cache but 'getInstance' actually returns a clone, not the instance itself.
Implemented by initializing one static instance in the view and reusing that instance between sorters.