Re: Newbie (Post#1): Help with Array of Objects

From: Java Architect (danc_at_dynamicresolve.com)
Date: 07/12/04 

Date: Sun, 11 Jul 2004 20:26:01 -0700

Difference of opinion and coding style. I have followed this rule since long
before I got into Java, and I get a result you failed to consider:

    5. The code is well written and runs successfully.

I code Java in a TDD environment (have for some time), and my tests will
catch cases 1-4 for me. Not giving an initial assignment to a variable leads
to problems like (this is contrived, but I've seen many similar cases):

int a;
try {
    a = 5;
    Thread.sleep(1000);
}
catch (InterruptedException err)
{
}
int y = a;

Oooops... You get an initialization error here.

You could argue that there's no reason to put the assignment in the
try/catch. In this simplified case, you'd be correct. However, there are
scenarios where you don't know the initialization value until somewhere in
the try/catch but before an unrelated (and inconsequential) exception might
be thrown.

"Chris Smith" <cdsmith@twu.net> wrote in message
news:MPG.1b5ba6bb14d620509896f9@news.altopia.net...
> Java Architect wrote:
> > This reply is exactly right, but there's more to it. In java, unlike
member
> > variables, local variables are not implicitly initialized, and you
cannot
> > use uninitialized variables except to assign a value to them.
>
> Right.
>
> > The point is that, when declaring a local variable, you should always
give
> > it an initial value.
>
> Yikes! That's about exactly the wrong conclusion. This only makes
> sense if you picture yourself in a war against the compiler to make your
> code compile, which is a dubious and even destructive world-view. If
> your goal is real working code, you'll cooperate with the compiler to
> help find bugs and make your code work properly. Ironically, it's
> exactly the opposite case in languages that don't have definite
> assignment checks (for example, C and C++); there you DO want to make
> sure you give every local variable an initial value.
>
> Look at it this way. There are several possible results to your making
> a mistake and accidentally using a variable that doesn't have a useful
> value. They are as follows:
>
> 1. The code always fails at build. Best possible scenario.
> 2. The code always fails at runtime. Good scenario.
> 3. The code can fail deterministically. Bad scenario.
> 4. The code can fail non-deterministically. Worst case scenario.
>
> In C and C++, the default is #4, but you can get to #3 by giving every
> local variable a value when it's declared, as a matter of policy.
> That's a good step, because it makes it easier to isolate and fix the
> problem once it's discovered. Sure you still won't find the problem
> until testing or quite possibly production, but there's no easy way to
> do any better, so you take what you can get.
>
> In Java, the default is #1, but you can get to #3 by giving every local
> variable a value when it's declared, as a matter of policy. That's a
> pretty bad direction to be going, because you WERE going to be told
> about the bug automatically as a matter of course, but you've just lied
> to the compiler to make it shut up, and now you won't find the bug until
> testing, or quite possibly production. You could lose customers over
> this kind of thing. The compiler is not your enemy; repeat as
> necessary.
>
> In practice, definite assignment checking helps you to find all kinds of
> issues, from actual incorrect treatment of that variable all the way to
> mistaken control flow, such as forgetting to abruptly terminate a method
> when responding to an exception. In any case, if you get that error and
> don't expect it, it's time to start looking for what's really wrong.
>
> There are, actually, a few situations in which definite assignment
> checking fails even though a variable is logically guaranteed to be
> assigned. For the most part, you can learn to avoid these by using the
> 'else' and 'default' keywords liberally. In a few situations, though,
> you really do need to assign a false initial value to avoid a compiler
> error. That indicates a pretty tricky piece of code, you you can
> simultaneously add the false initialization *and* a comment informing
> the reader about what's going on. That has never happened to me in
> seven years of writing professional Java code; but I can easily
> reproduce it in example code -- the point is that this should be
> exceedingly rare.
>
> So in the end, it's a really bad idea to always assign an initial value
> to a variable at its declaration. That's a nice advantage of Java; the
> right thing to do, at least in this case, is exactly what makes sense.
> No artificial rules required.
>
> --
> www.designacourse.com
> The Easiest Way to Train Anyone... Anywhere.
>
> Chris Smith - Lead Software Developer/Technical Trainer
> MindIQ Corporation

Relevant Pages

  • Re: Newbie (Post#1): Help with Array of Objects
    ... In java, unlike member ... > variables, local variables are not implicitly initialized, and you cannot ... sense if you picture yourself in a war against the compiler to make your ... The code always fails at runtime. ...
    (comp.lang.java.help)
  • Re: Cpp Considered Harmful
    ... >> programming language, the compiler, the IDE, the libraries, etc. ... source file, the implementation shall locate the declaration, (and ... to name one of the defining source files on the command line that initiates ... The counterpart to this in Java is accomplished using the following: ...
    (comp.lang.cpp)
  • Re: wie Array für statische Methoden
    ... > auf gar keinen Fall Java empfehlen;) ... Ich hab mal mit virtuellen Methoden auf einem Microkontroller experimentiert und bin zu dem Schluss gekommen: ... > Es ist ja gerade der Vorteil bei Java einen Compiler ... > Dinge zu optimieren, ...
    (de.comp.lang.java)
  • Re: What is the fastest method of parsing scheme?
    ... These issues can be eliminated by the use of custom memory allocators ... Any other ideas why Scheme would be faster than C++ and Java for heap ... For example, in my compiler, the procedure ) ...
    (comp.lang.scheme)
  • Re: A 21st Century Apple II?
    ... Java 6 -Xms64m 24.00s ... Which is actually 9.6% better than the C++ Intel compiler. ... I'm sure in some cases the GNU compiler produces better results. ...
    (comp.sys.apple2)