Favourite Mistakes

[NickPaul]

Here is one that I've made a number of times - I'll let you figure out what is wrong with it.

std::vector<ClassA> a;
std::vector<ClassA*> pa;
for(unsigned int i = 0; i < sum_num; i++){
   a.push_back(ClassA());
   pa.push_back(&a[i]);
}
//... 

I've got a few more good ones - I'll add them when I get a chance.

Please add any that you've been unlucky enough to come across.

Nick.

[NickPaul]

Just remembered another one:

struct StructA{
   unsigned char a;
   unsigned char b;
   StructA(unsigned char _a, unsigned char _b) : a(_a), b(_b) {}
};
bool operator<(const StructA& lhs, const StructA& rhs){
   if(lhs.a < rhs.a) { return true; }
   return false;
}

//...

std::map<StructA,int> some_map;
some_map[StructA(3,5)] = 3;
some_map[StructA(3,8)] = 8;

//...

[bluecoder]

what is the error you are getting here in the first one and in the second one .

[Disch]

The first one is flawed because vector can rearrange contents in memory when you append items.

So all the pointers being put in 'pa' might quickly become invalid pointers.

I didn't look much at the 2nd one.

[bartoli]

I'm guessing the second one is an incomplete comparison operator, so the two objects are inserted at the same position, the second erasing the first one?

[NickPaul]

@bartoli: Yep - that one took some serious debugging haha

[NickPaul]

This one got me good when I was getting started:

//...
unsigned int aSize = 10;
std::vector<StructA> a(aSize);
for(unsigned int i = aSize-1; i >= 0; i--){
   a[i].some_unsigned_int_member = i;
}
//... 

[kbw]

Didn't the compiler warn you?

[LB]

Nope. May I ask now, why the heck anyone would use postfix and not prefix ++ and -- operators? Don't give me any 'compiler optimization' stuff ;p

REAL men write code like this:

for(unsigned long i = 0; i < number; ++i) //preincrement ftw
{
    //
}

Anywho, I see a lot of people make this mistake with color codes:
color = (0, 255, 128);
"Why does it only have a halfway red color?"
Usually from people new to programming and trying to write HLSL shaders.

[Peter87]

L B wrote:

May I ask now, why the heck anyone would use postfix and not prefix ++ and -- operators?

A few years ago I thought prefix ++ was ugly. Now I don't know what I think so I use whatever I feel like.

[kbw]

They are different and can have different uses. Of course, if you just wnat to increment something, pre is faster. But the temporary used in the post can be handy.

[LB]

I understand there are uses, I just wonder why people would write i++ by itself instead of ++i

[Disch]

I think it's a holdover from C back in the day. postfix makes more sense in that context, as manually iterating with a pointer is more common.

I'd see this code a lot: *foo++;, which to this day I have to stop and think about exactly what it's doing (*stabs people who don't use parenthesis*).

Anyway postfix makes sense in that [relatively common] situation, and prefix didn't make sense in as many situations. And if you're using postfix in one place, why not use it in every place to be consistent?



Though I agree that prefix should be preferred unless you have reason to use postfix.