question 57 from Marcus Green mock exam

[Michael Pavlov]

Which of the following statements are true?
1) Adding more classes via import statements will cause a performance overhead, only import classes you actually use.
2) Under no circumstances can a class be defined with the private modifier
3) A inner class may under some circumstances be defined with the protected modifier
4) An interface cannot be instantiated
Right answers: 3 and 4
-------------------------------------------------
In my opinion 1) should be also true. I understood
Mr. Green explanation but consider this: I had a
big project with many authors before me; one of
them (not a real Java programmer considering his
code added in the import statements (of all
the code that he touched) all the possible classes
that were slightly close to the subject (probably just to make sure that everything is compiling fine and not to miss something, but he didn't care
that in fact half of the import statements waren't
necessary in the actual piece of code!). I was allowed( even encouraged by my mentor=guru, when he heard about it) to get rid of the unnecessary import statements and this made the code more robust actually! So quick question: what is your
opinion about this ?
Thanks in advance,
Michael

[Marilyn de Queiroz]

Some people have discussed the issue of imports and performance in the Performance forum.
http://www.javaranch.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic&f=15&t=000240
http://www.javaranch.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic&f=15&t=000305
http://www.javaranch.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic&f=15&t=000332
[ April 06, 2002: Message edited by: Marilyn deQueiroz ]

[Marcus Green]

Why would adding more classes via import statements cause a performance overhead?
Marcus

[Michael Pavlov]

Hi again,
I fallowed the links and I got what you said.
I was talking not only about wildcard imports of packages and the regular classes from the API (in which case you are perfectly right ofcourse!),
but especially about new classes being part of a
large application (like the one that I mentioned)
with more then 1000 java classes. We (my team)
observed that the product was faster meaning that
the time for execution was with 30-150 seconds
shorter then in the past ( the range is related
with what particular feature from within that product did you choosed to execute). That's all.
However, I'm not sure how this affected the compilation when all files get loaded, or the
runtime, but what I noticed is that the product
was a little bit faster after the elimination
of the import statements of classes not involved
in the actual piece of code...
Thank you very much for your input,
Best Regards,
Michael