Printing the stack trace vs the exception itself

Question

quick question. Of course the stack trace gives way more information, but is it a bad practice to use the exception itself instead in some cases? Like just getting "null pointer exception" as opposed to this huge dump of stuff? If this doesn't make any sense, the two differences would be:

(Exception e)
{
  print e
}

AND

(Exception e)
{
  e.printStackTrace
}

Answer 1

I will say use neither of them in production. You should handle exceptions in other ways, like logging the exception or saving the stacktrace somewhere to review it later and probably display a nice message to the user saying that the app or method failed.

But if you just ask about which of those approaches to use, I will say the latter is better because System.out.println(e) will call Throwable#toString that only provides the message of this exception and maybe you won't get the real cause of the problem.

Answer 2

It depends on the context. Anything that will (or even might accidentally) be presented to the user should not include the stack trace. More than being confusing, you could leak potentially sensitive information.

In my code, I create custom exceptions that have a UserFriendlyMessage property. Friendly messages should be in very plain English. I even have a default catch-all friendly message "There was an unexpected problem. You can try again and if the problem continues contact support".

On the other hand, having the stack trace is invaluable for debugging. In my .NET apps, we go so far as to compile our release code in a special way such that the code is optimized but the symbols are still generated in a separate file. We do this for the sole purpose of ensuring full stack traces with line numbers.

So in summary, display only friendly messages but log (and if you are going to, alert) full stack traces + the friendly message.