C Programming Puzzlers

by AndreyKarpov • 14 years ago

104 29 14 years ago

psykotic2• 14 years ago

Puzzles 9 and 12 assume sizeof(int) is 2. It's poor form for a language lawyering quiz to make an assumption about implementation-dependent behavior. It's especially poor form when the assumption made is long outdated and likely to surprise--I suppose it's still true for C compilers targeting 8-bit and 16-bit microcontrollers.

Aside: I remember buying the first Programmer's Heaven CD-ROM boxed set at The Party back in the mid-nineties. It was an incredibly valuable resource for me in those days. The best source of programming information was through the dial-up BBS scene, but none of the boards I frequented had a collection on the scale of Programmer's Heaven.

X-Istence• 14 years ago

From question 9:

"What is the output of this program on an implementation where int occupies 2 bytes?"

Question 12 is a function pointer question and doesn't say anything about ints at all.

I'll assume you meant Question 14:

"What is the output of this program on an implementation where int and all pointer types occupy 2 bytes?"

If you read the question you would have noticed that the assumption was put in writing so that even if you first assumed int was 4 bytes, or pointers were 4 bytes you would still be able to find the answer correctly.

psykotic2• 14 years ago

Ah, I had clicked through to the original article which does not contain those caveats.

Yes, I meant question 14, not 12.

finnw2• 14 years ago

Was it possible to initialise a local array with literals like that (e.g. Q5) in C89? If not then the version of C in the examples (assuming it is consistent) must be at least C99, which suggests that sizeof(int)==2 is due to the target being an embedded device rather than the program being old.

Edit: clarified that I am referring to local arrays only

hornd• 14 years ago

It was not only possible to initialize arrays like that in C89, that method is often referred to as "C89-style".

Also, let's not forget what the C standard says about integer sizes, which is that

A) It is guaranteed that sizeof char == 1

B) It is guaranteed that the following holds true: sizeof char <= sizeof short <= sizeof int <= sizeof long

So it's always completely unrealistic to assume anything about the size of these types on any system, except for the guarantee in (A)

bwarp3• 14 years ago

These are pretty much the reason that Go exists.

gue5t2• 14 years ago

These are not the problem with C. Things like a lack of support for functional programming constructs and more powerful manipulation of structs (e.g. iterating over fields and inspecting variable types) would be much more improvement to C than minor cosmetic changes that would hide the details that this quiz asks about.

vmmenon• 14 years ago

Lol.

bwarp3• 14 years ago

You've obviously know very little about the motivation of Go then...

Alind• 14 years ago

by saying this, I know you are not a progarmmer

bwarp3• 14 years ago

And you are a sanctimonious twat.

finnw2• 14 years ago

Edit: I was wrong - see cygx's reply

---

I'm a bit rusty on the C spec, but I think the author is wrong about #4.

The answer refers to an exception for a pointer that points one past the end of an array.

And &a[5] would be valid based on that rule, because it points one past the end of the array a.

However (&a + 1) is not valid. It does not point one past the end of an array (because although a is an array, it is not an element of an array.)

Having said that I would be surprised if any compiler gave a result other than "2 5". Still, it's technically undefined.

cygx2• 14 years ago

See C99 6.5.6 §7:

For the purposes of these operators, a pointer to an object that is not an element of an array behaves the same as a pointer to the first element of an array of length one with the type of the object as its element type.

brimpa• 14 years ago

For question 7, can someone confirm that

  int a[][3] = {1, 2, 3, 4, 5, 6};

is not a typo and is, in fact, a valid array initialization? I've never come across this and think it's rather unintuitive.

Sniffnoy• 14 years ago

Checking the rules for array initialization in Harbison & Steele, this does indeed seem to be invalid. I'm surprised, I actually thought that was valid...

cygx2• 14 years ago

It's valid (see C99 6.7.8 §20 and §22) and equivalent to

    int a[2][3] = { { 1, 2, 3 }, { 4, 5, 6 } };

grannyg00se• 14 years ago

Stopped at #3. It seems incorrect. The answer states that foo returns x to the power of n. But I ruled that answer out since any 0 or negative n is going to return 1.

edit: added emphasis

tzs• 14 years ago

Returning 1 for n = 0 is OK, as x^0 = 1 for all non-zero x, and 0^0 is usually also defined as 1.

However, as you note, it also returns 1 for negative n, and that is just wrong.

Well, wait a second....actually, if it were a machine where integer division rounds up[1], so that 1/k = 1 for k > 1, then #3 would correctly compute x^n for n < 0. Does this safe the author's answer?

Nope!

Their code for n > 0 would fail on such a machine. They are relying on repeated division by 2 eventually resulting in 0 in order to terminate the recursion. On a round up machine, repeated division converges to 1, not 0.

[1] I'm assuming that rounding behavior of integer division is implementation defined in C as defined at the time that quiz was written.

furyofantares3• 14 years ago

You could have ruled it out from the function signature alone:

    int foo(int x, int n);

This cannot be x^n, nor n^x or x*n.

raviksharma2• 14 years ago

"You could have ruled it out from the function signature alone" - kindly elaborate.

vainguard2• 14 years ago

In the answer to 9 I don't see why 'Evaluating ++i + ++i would produce undefined behavior'

prodigal_erik• 14 years ago

There's no http://en.wikipedia.org/wiki/Sequence_point between the two increments, so the standard allows the compiler to do pretty much anything, even http://catb.org/jargon/html/N/nasal-demons.html. The way I've heard it, these operators were added to take advantage of pre- and post-increment hardware support in the PDP-11 (forty years ago!), and everyone just knew what would happen, but ANSI later wanted compilers to have enough latitude to take advantage of other hardware which behaves differently.

vainguard2• 14 years ago

Just when I thought (after completing the puzzles) that C is reasonably reasonable language, I learn that the execution order of 'f()+g()' is unspecified. Wow!

C Programming Puzzlers

Related Stories