Pet Peeve

[Paulo Zemek]

I, for example, would never write the code like that. I can write an if without the {}, but I always put an extra line break after the call. And seeing two lines after the if simply looks wrong.

if ((err = SSLHashSHA1.update(&hashCtx, &serverRandom)) != 0)
goto fail;

if ((err = SSLHashSHA1.update(&hashCtx, &signedParams)) != 0)
goto fail;
goto fail; // why the hell there's a line of code here? It is not important what it does, it looks wrong.

if ((err = SSLHashSHA1.final(&hashCtx, &hashOut)) != 0)
goto fail;

So, this kind of error is as ugly to me as this (in fact, even uglier as the excessive {} somewhat hide the problem):

if ((err = SSLHashSHA1.update(&hashCtx, &serverRandom)) != 0)
{
goto fail;
}

if ((err = SSLHashSHA1.update(&hashCtx, &signedParams)) != 0)
{
goto fail;
}
goto fail; // Doesn't this look wrong?

if ((err = SSLHashSHA1.final(&hashCtx, &hashOut)) != 0)
{
goto fail;
}

But developers can also commit these errors (and by simply taking a looks without reading the code, I don't spot anything wrong):

if ((err = SSLHashSHA1.update(&hashCtx, &serverRandom)) != 0)
{
goto fail;
}

if ((err = SSLHashSHA1.update(&hashCtx, &signedParams)) != 0)
{
goto fail;
} // Oh, it looks like an else is missing...
{
goto fail;
}

if ((err = SSLHashSHA1.final(&hashCtx, &hashOut)) != 0)
{
goto fail;
}

// So the code gets corrected to...
if ((err = SSLHashSHA1.update(&hashCtx, &serverRandom)) != 0)
{
goto fail;
}

if ((err = SSLHashSHA1.update(&hashCtx, &signedParams)) != 0)
{
goto fail;
}
else
{
goto fail;
}

if ((err = SSLHashSHA1.final(&hashCtx, &hashOut)) != 0)
{
goto fail;
}

[robertschoenstein]

I personally prefer the former of the two, for readability. I tend to "speed read" classes and utilize the curly brackets as logical breaks in the code. Basically, for me, it differentiates between code indentations and logic (if, foreach, for, etc.)

ICP-Fan (The Keyboard Wielding Maniac)

[Paulo Zemek]

Imagine the if and the call are in the same line. Yet, the condition is false 99% of the time. Also, the method being called is called from many other places. If things are in two lines, simply put the breakpoint inside the if, not on the if line. If it is in the same line, you can't do that. Putting the breakpoint inside the method call will not help either, as it is called from many other places, so the best you can do is to create a conditional breakpoint. I prefer to have things in two lines and put the breakpoint exactly where it is needed.

[StatementTerminator]

Brady Kelly wrote:

I prefer

if (something) DoAB();

for the latter. I never use it, but have recently come across it. It seems more readable to me.

Me too, if you must omit the braces then it's best to put it on one line. That way you don't have to worry about someone coming along later, not paying attention, and doing this:

if (something)
DoA();
DoB();

[StatementTerminator]

Paulo Zemek wrote:

I agree with you. And I am actually the kind of person that when has to modify something like:

if (something)
{
DoA();
DoB();
}

To only call a DoAB(), I will go there and kill the extra { and }. So, I have more work doing that, but I keep consistency.
 
So, it becomes:

if (something)
DoAB();

Are you serious? You re-factor and combine methods just so you can avoid braces and use a dangerous bad practice?

[Paulo Zemek]

No, I don't refactor and combine methods... I was only giving an example. What's more probably is that I extract an entire block (with many calls) into a single method. But for the example I used two calls.

[StatementTerminator]

PIEBALDconsult wrote:

Quite. The maturity of a programming language may (in part) be calculated by how many newlines are required. A proper language requires none.

Just for making that statement, you should be forced to a maintain decade-old Perl CGI system.

[StatementTerminator]

Jeremy Falcon wrote:

Ok, you have a good point with that, but it's still not worth the trade off of having really extra long files / routines. Besides, it gives you something to spot the new guys with so you can give them a hard time.

Who cares how long the files are? They aren't being printed on paper (hopefully), and scrolling is not so hard. Setting a trap for junior programmers that introduces a bug is not responsible programming. If you do that, you are responsible for the bug, not the newbie, and you should be "given a hard time" for compromising the code base just to be a jerk.

[ronchristie]

I both agree and disagree. My personal coding style is:

if (expr) {
doSomething();
}
else {
doAnotherThing();
}

I like using the braces so that adding an extra bit of logic to the execution block won't inadvertently change the program flow, and at the same time placing the opening brace on the same line as the if... and the else statements compacts the code a bit. To me the advantage with compact code is that you can see more of the program logic in a single screen which means less time scrolling up and down trying to follow the program logic. For that reason, I'm ruthless in eliminating blank line whitespace except to visually offset functions or methods.

Ron Christie

[Jeremy Falcon]

The trap isn't that hard to spot though. Seriously man, we're not talking about something difficult. It's just a peeve. We can debate, but that's all it is. Just a peeve. And my preference is that I'd rather see more of the file, because it's not that hard to spot.

Jeremy Falcon

[Jeremy Falcon]

As much as I'm not a fan of that (even in JavaScript), that makes more sense than:

if (expr)
{
doSomethingForOneLine();
}
else
{
doAnotherThingForOneLine();
}

You're version is at least concise, and no matter what some may say on CP, concise is important. Less is more. Always will be.

Jeremy Falcon

[PIEBALDconsult]

Perl is a scripting language; not a programming language.

[Oleg A Lukin]

Well, I won't write anything such as that either. But since this a real world example of infamous 'goto fail' bug obviously other people do. In general I'd say that none of your samples with braces would've passed my code review, but I admit your first sample should attract review attention as well, it'll probably will attract enough attention from any reviewer. Still I'd say that with braces style the code like that most probably would not leave the developer and will be fixed before review

Banking establishments are more dangerous than standing armies. T.Jefferson

[StatementTerminator]

Yeah true, you really need to be pedantic about a joke?

[StatementTerminator]

LOL, that looks exactly like T-SQL :)

[PIEBALDconsult]

No, but I really need to be demeaning of Perl. X|

[rjmoses]

I personally prefer this convention: if ((a cond b) conjunction (c cond d) conjunction // Additional conditions always on separate line (e cond f)) { // Always include opening brace do .... something } // Always closing brace Or } else { // Always closing brace..else..opening brace do .... something else } // Always closing brace The same conventions and principles apply to loops (for, while, etc.) and other conditionals. I strictly avoid NOT using braces under any circumstances. I want to be able to see a complete thought, i.e., a block of code, in one intact, obvious section. These are my thoughts based on years of chasing bugs, modifying other people's code, and having to debug somebody else's problem. Other people are welcome to use their own conventions, but I may not be able to help them efficiently. Just because I can do something doesn't mean I should do it.

[jschell]

Karel Čapek wrote:

For example, how hard is it to

Both are probably less hard than complaining about it. But other than that since it is legitimate syntax then it is legitimate syntax and nothing but a preference after that. Naturally if that is the most important problem facing you day to day or even one that even rises above the noise level in terms of problems you must work at an exceeding stellar place. I once worked a a place where the VP (most senior person in the entire site) would regularly take down the database through incorrect usage which would shut down the call center (several hundred employees) completely. Now that is a problem.

[ronchristie]

It's interesting that the older the programmer the more concision becomes important. :)

Ron Christie

[Al Chak]

I have more

(condition)?DoThis():DoThat();

DoThese[(condition)]();

#define CALL_FUNC(__F) Do##__F
CALL_FUNC(condition);

do
{
if(condition)
{
DoThis();
break;
}
} while(DoThat()&&condition);

<pre lang="cs">
switch(condition)
{
case 1:
DoThis();
break;
default:
DoThat();
}

:laugh: