Difference between "throw(e)" and "throw e"?

Question

I came across this rethrown exception and am surprised that it even compiles.

} catch(SomeException e) {
    ...
    throw(e);
}

Is there any difference between this throw() and what is normally used?...

} catch(SomeException e) {
    ...
    throw e;
}

Any links to where this is documented or guidance on choosing one over the other?

Note that you shouldn't log and throw. If you catch it somewhere else and log it there too, you'll get the mistaken impression there were two exceptions. — Andy Turner, Dec 15 '17 at 21:49
There is a similar c++ question and the answer says no difference, https://stackoverflow.com/q/14321163 — Emre, Dec 15 '17 at 21:50
@AndyTurner depends. It's common that you want to log a problem somewhere where there is more context to the problem, but throw since you don't want to stop exception from propagating. — eis, Dec 15 '17 at 21:51
@eis In that case the context should be recorded in the exception itself. And while we're at it, logging `e.getMessage()` is a bad idea whether you rethrow or not. — biziclop, Dec 15 '17 at 21:51
I removed the log() line so as not to cloud the issue of the question. — noctonura, Dec 15 '17 at 21:53
..and when logging, you usually want the stack trace of the exception to be logged as well (`myLogger.error(e.getMessage(), e)`). `e.getMessage` often doesn't give enough information. — Mick Mnemonic, Dec 15 '17 at 21:53
@biziclop again, depends. It might be information that you do *not* want to expose to end user (where the exception might end up), but it is useful diagnostic in the logs. However I do agree that logging e.getMessage() is almost always wrong thing to do. — eis, Dec 15 '17 at 21:55
Basically `throw` accepts any expression that evaluates to an instance of `Throwable`. So you could even do `throw constructMyException();` if you've got a method `Exception constructMyException() {...}`. You can even do `throw null;`, but that's particularly naughty. And of course `(e)` is a valid expression and evaluates to ... well, `e`. — biziclop, Dec 15 '17 at 21:55
@eis you're talking about logging information that's not in the exception. I'm not objecting to that, I'm talking about not logging the exception and then throwing *the same* exception. — Andy Turner, Dec 15 '17 at 22:09

score 9 · Accepted Answer · answered Dec 15 '17 at 21:49

9

Quite a few languages allow as many parenthesis around expressions as you want. Java is one of them. The following is perfectly valid code.

public class HelloWorld {
  public static void main(String[] args) {
    throw ((((new RuntimeException()))));
  }
}

So there's absolutely no difference, except that your source file is two bytes larger.

answered Dec 15 '17 at 21:49

kelunik

6,750
2
41
70

2

"So there's absolutely no difference, except that your source file is two bytes larger." And much less readable. – davidxxx Dec 15 '17 at 21:56

score 6 · Answer 2 · answered Dec 15 '17 at 21:50

6

Functionally they are equivalent.

However, don't choose throw(e);, as someone might mistake it for a method call, and the very least will make someone unnecessarily wonder what it is that you're doing. Prefer the normal throw e; syntax for clarity.

answered Dec 15 '17 at 21:50

eis

51,991
13
150
199

2

Right. Unnecessary parenthesis that are neither functionally required nor a readability improvement are error prone and dead code. – davidxxx Dec 15 '17 at 22:07

score 0 · Answer 3 · answered Dec 15 '17 at 22:20

Throw is an instruction to throw a "throwable" (usually an exception)

Think if it like a return statement

Public int get value() {
    return 3;
}

Is equlivilant of

Public int get value() {
    return (3);
}

It's the same with throwable.

In face they will complied to the exactly same thing.

Difference between "throw(e)" and "throw e"?

3 Answers3