- Feature Articles
- CodeSOD
- Error'd
-
Forums
-
Other Articles
- Random Article
- Alex's Soapbox
- Announcements
- Best of…
- Best of Email
- Best of the Sidebar
- Bring Your Own Code
- Coded Smorgasbord
- Mandatory Fun Day
- Off Topic
- Representative Line
- News Roundup
- Editor's Soapbox
- Software on the Rocks
- Souvenir Potpourri
- Sponsor Post
- Tales from the Interview
- The Daily WTF: Live
- Virtudyne
Admin
My friend Sheila is particularly dismissive of that "rule", for some reason. Weird.
Admin
What about
return (condition1(x) || !condition2(x) || condition3(x));
Admin
Yep, if authCode is null you are in all sorts of trouble.
Whereas "".equals(null); still works.
Admin
It's part of my company's coding standards, and I like my job, so I do it. Someone said earlier that it's a habit. Actually, it's a preference. So is using multiple returns. I doubt that one is better than the other, but if there's an industry standard, there's generally a reason for it, and it's probably better to code to that standard, if only so that your code is seen to be "professional" by such non-techie types as managers and human resources reps... particularly those who view samples of your work during interviews.
Admin
incorretly returns true if all conditions evaluate to false; incorrectly evaluates codition3 if condition1(x) == false and condition2(x)==true
Admin
should then simply be
return (condition1(x) || !condition2(x)) && condition3(x));
Admin
ERRATA
okay, I get out, it doesn't work better :)
so, i'm going to spend some minutes to find out a way =)
Admin
c1 || (!c2 && c3)
that's marvelous
I stop flooding here, thanks for listening :p
Admin
If you're using the _com_ptr_t class (or maybe CComPtr too, I've never used that), there's no need for the !!. They define an operator bool, which allows for automatic conversions to a bool for if statements and such. If it's your own class, maybe put in something like
//first post here, let's hope it doesn't get fubared...
Admin
kipthegreat: "Opera is for hippies, gays, and terrorists. Use Firefox. It's what the cool kids are using."
In /this/ place, I can't tell if that's biting satire of a far too large section of the rabit Firefox fanboy community, or an actual Firefox fanboy.
Either way, I suppose I'd better live up to this hippy, gay, terrorist slander and kill you with an exploding hamster of peace.
Admin
Good- but for full points, I'm afraid, you've got to throw in the ternary operator:
becomes:
Also- I'd prefer to see "004" and "005" be made into parameters- preferably enums.
Admin
What a bunch of vagina's, the code's not right, wa, wa, I am of course assuming it works, not all code in "perfect" to everyone's standards, in fact I would bet less than 20% would pass a large group of nerds review.
So you could do it better, we all could, but it works and get over it! Now you would certainly point out how that's not the way to do it.
It is truly a WTF though!
Admin
Why, I wrote stuff like that. With reason: cond is often a function that returns BOOL, as a magnitude of stuff does in MFC. BOOL is a typedef for int, not bool, and when it is converted to the bool return type of the function the compiler will give a warning like 'performance whatever: forcing int to bool'
You may be in a situation where turning off that warning is not preferred -- but !! eliminate it for good while the generated code is identical.
Admin
You're right - who needs code clarity when you can save 5 bytes of precious hard disk space?
Admin
Admin
Though you have to be careful not to write
if(a)
if(b) f();
else g();
thinking you're being cool for not using braces.
Or get really cocky and start putting things all on one line.
lock(); stuff(); unlock();
if(a) lock(); stuff(); unlock();
Admin
Oops, this is in reply to an old post that said:
Admin
Admin
Yeah but nobody said "Who.is01()?"
Admin
Well, there's some good news: at least all the comparisons have the literal on the left to avoid Null Pointer Exceptions.
Further good news: this is easily and mechanically refactored to something potentially useful.
I can't figure out how to spin this positively beyond that. Sorry.
Admin
if (cond)
{
retVal = a;
goto exitfunction;
}
retVal = b;
exitfunction:
return retVal;
Admin
The "one return value" is a throwback to when "structured programming" became fashionable in the 80s
Admin
I agree that the single-return-point has its merits. I still believe the drawbacks outweigh the gains.
When you have multiple natural points, but are forced into the single return idiom, you're going to run out of room to indent:
if (!cond_a)
{
....
if (!cond_b)
{
...
if (!cond_c)
{
// getting ridiculous
}
}
}
return retval;
That
Admin
...drives me nuts (guess I shouldn't accidentally hit the "HTML" tab mid-post and try to flip back to "Design").
Admin
surely they meant
private <FONT color=#0000ff>String</FONT> isBrillant(String authCode)
{
if ("Brillant".equals(authCode)) return <FONT color=#0000ff>"Paula"</FONT>;
}
Admin
or when sounded like "A" (as in neighbor and weigh)
or the case where a word ending in e does not drop the final vowel when forming the continuous present tense (as in seeing)
or the few special cases that do not follow any rule (mostly stuff imported from another language)!
Admin
Yes. Your version throws NullPointerExceptions. The original does not.
Admin
"Exploding Hamsters of Peace" would make a great name for a band.
Admin
You use goto? Expect hellfire and brimstone to rain down on you at any moment...
Admin
say you have functions condition[n](x) that modify the value of x and return a boolean.
if (condition1(x)) return true;
if (condition2(x)) return false;
if (condition3(x)) return true;
return false;
if (condition1(x)) {
retval = true;
} else if (condition2(x)) {
retval = false;
} else if (condition1(x)) {
retval = true;
} else {
retval = false;
}
return retval;
return condition1(x) || !condition2(x) || condition3(x);
that's a great way to do it and it will work perfectly. that is... assuming your compiler/language uses short circuit boolean evaluation of the operator ||
Admin
I just realized if c1 is false and c2 is true you have an error because c3 gets excecuted and you didn't want c3 to get run at all.
Admin
You don't seem all that knowledgeable about compilation.
Admin
Why not use:
Admin
" You don't seem all that knowledgeable about compilation."
I most certainly am not a compiler expert.
Admin
I'm getting the distinct feeling not everyone read all the posts before replying (as I do)...
Btw, isn't this thread flooded with more logically incorrect code than normally? WTF???
Admin
Hah, sure, returns
trueifcondition2(x)is false instead of evaluatingcondition3, and it returnstrueifcondition2(x)istrueandcondition3(x)is true while it should returnfalseFucking great refactoring.
See above
Admin
All of this rather proves the point: the original version with multiple returns is far more readable and maintainable, for the simple reason that it's how people think.
Admin
If Condition1 is true, that expression will go on to evaluate Condition3, which the original did not, and if Condition3 is false ther answer will be different, so again it is incorrect.
As I said, the original with multiple return points is far easier to read and understand, thus blowing this "law" out of the water.
Admin
Learn to quote, Anthony. I didn't write that.
Admin
Why not just change that to...
if (cond)
{
// Do something useful...
retVal = a;
}
else
{
// Do something else useful...
retVal = b;
}
return retVal;
...instead? You get the single return point you want, plus the added bonus of improved readability without having to unleash the blinding, unholy power of a goto.
Between you and me, I think else is really just an labelless goto in disguise. But don't tell anyone, ok?
Admin
Hello (im fresh here, writin' the very first time ,-)
The statement "return !!cond;" is used to normalize the result to false and true, whatever the input value "cond" ist. One sees this is coding from C.
It's farly short, but in my opinion and technically all right!
floh
Admin
You should enlighten us all with your vast knowledge.
Admin
It would be easier to write some code and look at the assembly code that comes out of the compiler.
Some people come here, I hope, to learn what not to do when programming, so I think it's appropriate to point out when someone recommends coding a particular way based on basically wrong assumptions about how compilers work -- such as the assumption that programs somehow have to "pass through" a branch that isn't taken.
Admin
Quite.
Another thing is that there are some people who know a more about certain areas. This is more valuable than ignorance. The poster two up should consider this instead of blabbing. The attitude that someone who knows should be scorned for knowing leads to WTFs. In this case, the knowledge about compilers being referred to is insultingly basic and should be known by anyone using a compiler. No vast knowledge is needed.
Sincerely,
Gene Wirchenko
Admin
Actually, some CPU architectures (e.g. ARM) allow conditional execution of most statements (not only jumps), so it's quite possible that the program passes through the branch that isn't taken, though it doesn't do anything there. That's because a jump is relative expensive, especially without branch prediction.
Admin
I wouldn't go that far. It should be known by anyone making performance suggestions.
And yeah, ammoQ showed me up a little. But that's a case where the compiler sends the program through the alternate branch to improve performance, not because it's not smart enough to skip over.
Admin
When you put the condition and the resulting action on the same line, in most IDE's you cannot set a breakpoint on the action alone, the debugger always breaks regardless of the value of the conditon. This is not always what I want.
Admin
Then look at your passport! You ARE in fact an ID to your state! ;P
(A resource-id to be exactly. Only in use as long as of use.)
----
P.S.: CAPTCHA WTF? Is "boko" a word?? Or is it "boo"?
Can Alex please explain for us all, why he's still using that poor forum software despite the many complains he gets about it? Is it that there is a special features that makes it rock? Or is it the "Internet Explorer User" syndrome? (The one where poeple think they can't let loose from IE because it is "faster" or has this one special feature FF has not (ignoring that there is an extension for it or that it is simply irrelevant when you compare it to the consequences of using IE.))
P.P.S.: Okay. As always, the first captcha does not work... Seriously... Quad-WTF?
P.P.P.S: Okay. I know i wrote it correctly because it was pretty easy. But again it did not work. No comment! This time it is "china" just for the log...
Admin
Well guys, believe it or not: I've worked on this project and it's for real! Better still: this stuff is used to calculate our PENSIONS and other social security!!! So I better start putting my money in a sock or matres :-(
Cheers,
Admin
If you don't allow more than one return statement and obviosly no gotos, then why allow stuff like break or continue?
Consider the following:
for (unsigned i = 0; i< L ; ++i) {
if (someCondition(array[i])) break;
doStuff;
}
to stop when the conidition is met.
now try it with nested loops:
for (unsigned i = 0; i< L ; ++i) {
for (unsigned j = 0; j< L ; ++j) {
if (someCondition(array[i][j])) break;
doStuff;
}
}
The break only stops the inner loop. A goto to the end might be smart here if you want to continue or a return in case it's the only thing to do:
bool func () {
for (unsigned i = 0; i< L ; ++i) {
if (someCondition(array[i])) return true;
doStuff;
}
// condition wasn't met
return false;
}
Now if you want to use that "retval", you'd have to make something like:
bool func () {
bool retVal = false;
for (unsigned i = 0; i< L ; ++i) {
if (someCondition(array[i])) {
retVal = true;
.... break? goto?
}
doStuff;
}
return retVal;
}
or use something like:
bool func () {
bool retVal = false;
for (unsigned i = 0, bool end_loop = false ; i< L || loop_end ; ++i) {
if (someCondition(array[i])) {
retVal = true;
end_loop = true;
... continue? // surely it's better than trying to skip doStuff with a conditional.
}
doStuff;
}
return retVal;
}
-----------------------------------------
Another example... many times you want to do something only if certain pre-conditions are met, or optimizing by "short-circuiting", for example:
bool func() {
if (!condition1) return false;
doComputationRequiringCondition1;
if (!condition2) return false;
doComputationRequiringCondition2;
// all conditions met, true!
return true;
}
And for those who might have thought otherwise, at least in C, C++, java, C#, perl the following happens for this line of code:
if (f() && g()) { doSomething; }
f() is first called,
if it evaluates to false, it continues to the next line, without evaluating g().
if it evaluates to true, g() is called
if it evaluates to true, "doSomething" will occur
if it evaluates to false, it continues to the next line.
so assume cond2 requires cond1 for its calculation (for example, cond1 checks if x is positive and cond2 takes the square root of cond1....) then writing:
if (cond1(x) && cond2(x))...... is completely logical and cond2 will NOT run if cond1 fails.