🚀 go-pugleaf

Good C interview questions? Consider asking what this function does and why:

char	xxx(int n)
{
	return "0123456789ABCDEF"[n];
}

Later,
 AriL
-- 
All my opinions are mine and mine alone.

Ari Lukumies <aril@cmt.lpr.mail.carel.fi> wrote:

>Good C interview questions? Consider asking what this function does and why:

>char	xxx(int n)
>{
>	return "0123456789ABCDEF"[n];
>}

>Later,
> AriL
>-- 
>All my opinions are mine and mine alone.

consider n > 15
Andrew Snyder
a1s@ix.netcom.com
char disclaimer[] = {'\0'}; /* I'm on cash net */

a1s@ix.netcom.com (Andrew Snyder) wrote:

>Ari Lukumies <aril@cmt.lpr.mail.carel.fi> wrote:

>>Good C interview questions? Consider asking what this function does and why:

>>char	xxx(int n)
>>{
>>	return "0123456789ABCDEF"[n];
>>}


>consider n > 15

consider n < 0

Interview question: a) what does this function do and why?
b) rewrite the function to handle arbitrary input.

[The correct answer to b) is, of course, "what should I return for
 values out of range?" -mod]

==
Miguel Carrasquer Vidal                     ~ ~
Amsterdam                   _____________  ~ ~
mcv@pi.net                 |_____________|||

========================== Ce .sig n'est pas une .cig

Ari Lukumies (aril@cmt.lpr.mail.carel.fi) wrote:

> char	xxx(int n)
> {
> 	return "0123456789ABCDEF"[n];
> }

My personal favorite:

Given:
	char*  A = "hello" ;
	int    i =  3 ;

It's obvious that:

	A [i] == 'l'

But what is:

	i [A] == ???

Any competent C programmer should know.

--
Chris Sonnack  <cjsonnack@mmm.com>                  http://eishcq.mmm.com
Engineering Information Services/Information Technology/3M, St.Paul, Minn
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
An intellectual is someone whose mind watches itself

In article <4hqebv$1k1@solutions.solon.com>,
Chris Sonnack <cjsonnack@mmm.com> wrote:
>My personal favorite:
>
>Given:
>	char*  A = "hello" ;
>	int    i =  3 ;
>
>It's obvious that:
>
>	A [i] == 'l'
>
>But what is:
>
>	i [A] == ???

>Any competent C programmer should know.

Any *competent* C programmer should know ??? Come on, have you *ever*
seen this weird construct used outside of the IOCCC ?
(Maybe this is a serious question... As for me, I don't remember having
ever seen it elsewhere).

What is its incidence on reality, as defined by programs written to accomplish
something and not to obfuscate ? Knowing the symetry of C arrays is fun,
but there are loads and loads of more useful things that a C programmer
should know first and foremost. Why not consider trigraph knowledge a must
as well ?

-- 
[nosave]<http://www.eleves.ens.fr:8080/home/espie/index.html>
microsoft network is EXPLICITLY forbidden to redistribute this message.
`Moon purismu powa, make up.... Tsuki ni kawatte, oshiokiyo !'
	Marc Espie (Marc.Espie@ens.fr)

In article <4hv081$jdt@solutions.solon.com>,
Marc Espie <espie@clipper.ens.fr> wrote:

>>It's obvious that:
>>	A [i] == 'l'
>>But what is:
>>	i [A] == ???
>
>>Any competent C programmer should know.
>
>Any *competent* C programmer should know ??? Come on, have you *ever*
>seen this weird construct used outside of the IOCCC ?

I've used it once.  In the process I discovered that IBM's xlc compiler
doesn't like reverse array notation -- it generates bad code.

I think a C programmer should know that array indexing is commutative,
but I wouldn't ask about it in an interview.  Unless I was interviewing
somebody just out of school for a first real programming job I wouldn't
ask detailed C syntax/semantic questions at all.  There are too many
details of C.  It would take a long time to ask about them all, and it
is reasonable not to know some of them.

Q: Which has higher precedence, + or |?
A: I don't remember, so I use parentheses.

I really didn't remember and I've been using C for more than 10 years.
When I need to know I take 30 seconds to write a C program to see what the
compiler does, or read the standard, or look at gcc's parser.  (Sun's C
compiler says + is higher than |.)  Maybe that would be a good interview
question: ask what the precedence is and if the answer is "I don't know"
follow up by asking: "show me how you would find out".

-- 
    John Carr (jfc@mit.edu)

John Carr wrote:
> 
(snip discussion on reverse array notation)
> 
> Q: Which has higher precedence, + or |?
> A: I don't remember, so I use parentheses.
> 
> I really didn't remember and I've been using C for more than 10 years.
> When I need to know I take 30 seconds to write a C program to see what the
> compiler does, or read the standard, or look at gcc's parser.  (Sun's C
> compiler says + is higher than |.)  Maybe that would be a good interview
> question: ask what the precedence is and if the answer is "I don't know"
> follow up by asking: "show me how you would find out".
> 
> --
>     John Carr (jfc@mit.edu)

I agree completely! Nobody can remember everything, but knowing how to
answer your own questions is probably one of the most important skills a
prospective employee can have. (One of the others is knowing what to
ask.)
-- 
 -----------------------------------------------------------
| Eric Goebelbecker    |                                    |
| Market Vision        |   You are a strange,               |
| 40 Rector Street     |              sick, little man."    |
| New York, NY. 10006  |                                    | 
| phone: (212)306-0355 |                 - Buzz Lightyear   |
 -----------------------------------------------------------

In article <4hv081$jdt@solutions.solon.com>, espie@clipper.ens.fr (Marc Espie) writes:
|> In article <4hqebv$1k1@solutions.solon.com>,
|> Chris Sonnack <cjsonnack@mmm.com> wrote:
|> >My personal favorite:
|> >
|> >Given:
|> >	char*  A = "hello" ;
|> >	int    i =  3 ;
|> >
|> >It's obvious that:
|> >
|> >	A [i] == 'l'
|> >
|> >But what is:
|> >
|> >	i [A] == ???
|> 
|> >Any competent C programmer should know.
|> 
|> Any *competent* C programmer should know ??? Come on, have you *ever*
|> seen this weird construct used outside of the IOCCC ?
|> (Maybe this is a serious question... As for me, I don't remember having
|> ever seen it elsewhere).

It also appears in test suites for compilers.  IMO your overall point
is well taken, however.

|> What is its incidence on reality, as defined by programs written to accomplish
|> something and not to obfuscate ? Knowing the symetry of C arrays is fun,
|> but there are loads and loads of more useful things that a C programmer
|> should know first and foremost. Why not consider trigraph knowledge a must
|> as well ?

jfc@mit.edu (John Carr) writes:

>In article <4hv081$jdt@solutions.solon.com>,
>Marc Espie <espie@clipper.ens.fr> wrote:

>>>It's obvious that:
>>>	A [i] == 'l'
>>>But what is:
>>>	i [A] == ???
>>
>>>Any competent C programmer should know.
>>
>>Any *competent* C programmer should know ??? Come on, have you *ever*
>>seen this weird construct used outside of the IOCCC ?
 
Does IOCCC stand for "International Organization of Competent C Coders"?  
 
Since A[i] is turned into *(A+i), I was able to figure out on my own 
that if the second expression at the top of this post compiled, then it 
would have to produce the same result as the first one.  However, I would 
have thought that the compiler would do a check whenever it saw brackets 
to make sure that the thing before the brackets was a pointer and the 
thing inside the brackets was an integer.  However, K&R state that 
indexing is commutative, and a quick test program compiled and run 
using my (admittedly substandard) compiler confirms this.  So I learned 
something new today!  :)  

>I've used it once.  In the process I discovered that IBM's xlc compiler
>doesn't like reverse array notation -- it generates bad code.

>I think a C programmer should know that array indexing is commutative,
>but I wouldn't ask about it in an interview.  Unless I was interviewing
>somebody just out of school for a first real programming job I wouldn't
>ask detailed C syntax/semantic questions at all.  There are too many
>details of C.  It would take a long time to ask about them all, and it
>is reasonable not to know some of them.

>Q: Which has higher precedence, + or |?
>A: I don't remember, so I use parentheses.

>I really didn't remember and I've been using C for more than 10 years.
>When I need to know I take 30 seconds to write a C program to see what the
>compiler does, or read the standard, or look at gcc's parser.  (Sun's C
>compiler says + is higher than |.)  Maybe that would be a good interview
>question: ask what the precedence is and if the answer is "I don't know"
>follow up by asking: "show me how you would find out".

My answer would be:  I would just look on page 53 of Kernighan & Ritchie.  
(Since I already know said page number by heart.)  
 
                       - Kevin Jay Miller 
 ` 

John Carr wrote:
> 
(snip discussion on reverse array notation)
> 
> Q: Which has higher precedence, + or |?
> A: I don't remember, so I use parentheses.
> 
(snip comment and agreement)

I have to wonder, in what context would you want to mix arithmetic
and bitwise operations?  It strikes me that, with modern optimizing
compilers, you will save NOTHING (except clarity :-]) by doing this.

My answer would be, 
A: Who cares? The code is either flaky or unclear.


Regards
James McLeod
mcleod@ecf.utoronto.ca

Marc Espie (espie@clipper.ens.fr) wrote:
: In article <4hqebv$1k1@solutions.solon.com>,
: Chris Sonnack <cjsonnack@mmm.com> wrote:
: >My personal favorite:
: >
: >Given:
: >	char*  A = "hello" ;
: >	int    i =  3 ;
: >
: >It's obvious that:
: >
: >	A [i] == 'l'
: >
: >But what is:
: >
: >	i [A] == ???
: 
: >Any competent C programmer should know.
: 
: Any *competent* C programmer should know ??? Come on, have you *ever*
snip
: What is its incidence on reality, as defined by programs written to accomplish
: something and not to obfuscate ? Knowing the symetry of C arrays is fun,
: but there are loads and loads of more useful things that a C programmer

Nope, never seen it in practice.  But, I did have it drilled into my head
during my advanced C class.  The point is not that he knows something
obfuscated but, rather, he understands what the compiler is doing with
the given construct.  It really has more to do with the equivalency of
arrays and pointers than commutative properties.  If the interviewee can
answer the question then you know there is a good chance that he has
discussed things like a[i][j] == *( *( a + i ) + j ).  The fact that
the FAQ for this group has an entire chapter devoted to such a topic
simple underscores its relevance.

 >: Chris Sonnack <cjsonnack@mmm.com> wrote:
 >: >My personal favorite:
 >: >
 >: >Given:
 >: >	char*  A = "hello" ;
 >: >	int    i =  3 ;
 >: >
 >: >It's obvious that:
 >: >
 >: >	A [i] == 'l'
 >: >
 >: >But what is:
 >: >
 >: >	i [A] == ???

What's confusing about this?  The "i'th element of A" is
about as intuitive as you can get.  Only frustrated fortran
programmers would prefer the former :-).

Mike

Michael J. Creutz wrote:
 > 
 >  >: Chris Sonnack <cjsonnack@mmm.com> wrote:
 >  >: >My personal favorite:
 >  >: >
 >  >: >Given:
 >  >: >   char*  A = "hello" ;
 >  >: >   int    i =  3 ;
 >  >: >
 >  >: >It's obvious that:
 >  >: >
 >  >: >   A [i] == 'l'
 >  >: >
 >  >: >But what is:
 >  >: >
 >  >: >   i [A] == ???
 > 
 > What's confusing about this?  The "i'th element of A" is
 > about as intuitive as you can get.  Only frustrated fortran
 > programmers would prefer the former :-).
 > 

You mean that i[A] is preferred to A[i] ? I don't agree. The standard makes the 
two identical, but I wouldn't want to maintain code using the latter approach. 
Well, maybe if you were coding Lisp...

Later,
 AriL
-- 
All my opinions are mine and mine alone.

I've read a lot of posts in this thread about what _would_ be good
interview questions.  Well, I have a few that _were_ asked of me when
I had a telephone interview with SGI a few weeks ago.

What does it mean when you declare something static?
  (I supplied answers for file scope, local variables, and inside
   C++ classes.  Do you guys know of other situations?)

What does the comma operator do, and what does it return?

What's the best way to get the size of this structure?

   struct foo {
      int a;
      char b;
      int c;
      char d;
   };

  (I said, "sizeof(struct foo)".  He chuckled and said, "I can tell you're
   a C programmer.  A C++ programmer would have just said sizeof(foo)." :-))

   Pick your favorite architecture.  What would the sizeof(foo) be on it?

  (I picked what we have at school, Sun workstations with gcc.  Should have
   said SGI's :-))

   Why isn't the size 2*sizeof(int)+2*sizeof(char)?

   How could you decrease the size of the structure without removing
   any members?

Then some C++ <crap> on which I didn't as well, but you guys wouldn't
wouldn't want to hear about that...

Then some data structures stuff, like running time for quick sort, heap
sort, hash tables, and various operations on singly and double linked
lists.  Then, he asks, "OK, you've correctly told me that some operations
are more efficient for doubly linked lists than for singly linked lists,
but doubly linked lists use the extra memory of a pointer at each node.
How can we get all the efficiency of a doubly linked list, but use
only as much memory as a singly linked list, assuming the full significance
of the pointers is required?

This one, I heard about once on this newsgroup.  Store the bitwise XOR of the
addresses of the next and previous nodes in each node.  BTW, is there
another answer you guys know of?  I have some sample code that implements
this for any interested parties, email me if you want it.

--
Ed L. Karrels
karree46@sol.acs.uwosh.edu
http://www.uwosh.edu/students/karree46/

In article <4jibmd$8b5@solutions.solon.com> <Ed.L.Karrels@zippo.com> writes:
>... I have a few that _were_ asked of me [recently]....  Then, he
>asks, "OK, you've correctly told me that some operations are more
>efficient for doubly linked lists than for singly linked lists,
>but doubly linked lists use the extra memory of a pointer at each node.
>How can we get all the efficiency of a doubly linked list, but use
>only as much memory as a singly linked list, assuming the full significance
>of the pointers is required?

>This one, I heard about once on this newsgroup.  Store the bitwise XOR
>of the addresses of the next and previous nodes in each node.

It is probably worth pointing out that, while this trick works
quite often, it is ultimately not portable and should generally be
avoided.  (Your first problem is finding an integral type long
enough to hold a pointer on every implementation.  On some, there
may not be any such type, and you might even have to resort to an
array of `unsigned char'!)

Indeed, if space is *that* tight, you may be better off using an
array of elements, doing the linking with array indices.  That is,
instead of:

	struct elem {
		struct elem *forw, *back;
		...
	};

	void init_queue(struct elem *headp) {
		headp->forw = headp->back = headp;
	}

	/* insert element `e' after element `after' */
	void insert(struct elem *e, struct elem *after) {
		e->forw = after->forw;
		e->back = after;
		after->forw = e;
		e->forw->back = e;
	}

you might have as your initial transformation:

	struct elem {
		unsigned int forw, back;
		...
	};

	struct elem space[MAX_ELEM];
	#define qptoix(qp) ((qp) - &space[0])
	#define ixtoqp(ix) (&space[ix])

	void init_queue(struct elem *headp) {
		headp->forw = headp->back = qptoix(headp);
	}

	/* insert element `e' after element `after' */
	void insert(struct elem *e, struct elem *after) {
		e->forw = after->forw;
		e->back = qptoix(after);
		after->forw = qptoix(e);
		ixtoqp(e->forw)->back = qptoix(e);
	}

Now two indices *can* be xor'd portably, and the result uses the
same number of bits as the original values.  If the array size is
65536 or smaller, an index can be stored in an unsigned short.  On
a machine with 64-bit pointers, this could save 14 bytes per element.
(Of course, we might wonder why a machine with 64-bit pointers was
so tight on space in the first place. :-) )
-- 
In-Real-Life: Chris Torek, Berkeley Software Design Inc
El Cerrito, CA	Domain:	torek@bsdi.com	+1 510 234 3167

Ed.L.Karrels@zippo.com inexplicably wrote:
} I've read a lot of posts in this thread about what _would_ be good
} interview questions.  Well, I have a few that _were_ asked of me when
} I had a telephone interview with SGI a few weeks ago.

} What does it mean when you declare something static?
}   (I supplied answers for file scope, local variables, and inside
}    C++ classes.  Do you guys know of other situations?)

} What does the comma operator do, and what does it return?

} What's the best way to get the size of this structure?

}    struct foo {
}       int a;
}       char b;
}       int c;
}       char d;
}    };

}   (I said, "sizeof(struct foo)".  He chuckled and said, "I can tell you're
}    a C programmer.  A C++ programmer would have just said sizeof(foo)." :-))

}    Pick your favorite architecture.  What would the sizeof(foo) be on it?

A compile error, right?

}   (I picked what we have at school, Sun workstations with gcc.  Should have
}    said SGI's :-))

}    Why isn't the size 2*sizeof(int)+2*sizeof(char)?

Byte alignment?

}    How could you decrease the size of the structure without removing
}    any members?

Rearrange the elements.

} Then some C++ <crap> on which I didn't as well, but you guys wouldn't
} wouldn't want to hear about that...

} Then some data structures stuff, like running time for quick sort, heap
} sort, hash tables, and various operations on singly and double linked
} lists.  Then, he asks, "OK, you've correctly told me that some operations
} are more efficient for doubly linked lists than for singly linked lists,
} but doubly linked lists use the extra memory of a pointer at each node.
} How can we get all the efficiency of a doubly linked list, but use
} only as much memory as a singly linked list, assuming the full significance
} of the pointers is required?

} This one, I heard about once on this newsgroup.  Store the bitwise XOR of the
} addresses of the next and previous nodes in each node.  BTW, is there
} another answer you guys know of?  I have some sample code that implements
} this for any interested parties, email me if you want it.

Could you maybe post it?  What about it, moderator?

[I'd rather see a pointer to it posted, unless it's quite small.  In no
 case would I like to see a URL for this code, and the code, xor'd together
 and posted.  See also Torek's post about it, just recently approved.  -mod]
--
Bill Wendling            bwendlin@fgi.net 
GCS/M/S d s-:+ a? C++++$ UL++>++++ P+++>+++++ L+++>+++++ E--- W-- N++ !o
!K w--- O M-- !V PS+++@ PE- Y !PGP t+++(*) 5-- X- R++ tv--- b+++ DI++++
!D G+ e++>+++++ h(++) !r !y+

Ed.L.Karrels@zippo.com wrote:

>I've read a lot of posts in this thread about what _would_ be good
>interview questions.  Well, I have a few that _were_ asked of me when
>I had a telephone interview with SGI a few weeks ago.

[questions and answers omitted]

Well, did you get the job?


==
Miguel Carrasquer Vidal                     ~ ~
Amsterdam                   _____________  ~ ~
mcv@pi.net                 |_____________|||

========================== Ce .sig n'est pas une .cig