log will make a 5 times degradation in performance!
so if your call depth is 5, and you log at every layer, then that single failed call will take 5**5 more time to complete!
not to say the big mess made by all the repeated log record for the same single failure.
so in the end, log will kill you instead of save you. :-)
John Xie
SourceCompanion, the best in class collabration tool built on vss
catch and rethrow is expensive, log is far far more expensive!
Rodz
i think i was replying in another thread.
the original talk was about exception management.
when I refer to log, i mean trace log.
the cause of this overhead is due to the way tracelog is designed, the lock on concurrent access.
lordmarc