Kazu profiled his code and found that using the current haskell date/time libraries was extremely expensive. He solved the issues by writing optimized date code. It is unclear how much benefit there would be to caching after switching to his optimized date code. He is planning to release his logging code as a separate package.
Did you measure how long it takes to generate that datetime string? Shouldn't that be possible in a hundred nanoseconds or so?
Am 08.08.2011 15:01 schrieb "Gregory Collins" <greg@gregorycollins.net>: