Use of nested loops.

[Ben Pfaff]

Hmmm. How about:

Larry meant what he said. Read C99 6.8.5:

5 An iteration statement is a block whose scope is a strict
^^^^^^^^^^
subset of the scope of its enclosing block. The loop body is
also a block whose scope is a strict subset of the scope of
the iteration statement.

The corresponding section in C90 doesn't include anything similar.

 

[Morris Dovey]

I count 6 block scopes. In C99 you could have declared i, j, and k
within the initialization statement of the for loop. IIRC, the body
of a loop opens a new block scope. This does not necessitate the
use of curly brackets.

I could have, but didn't. I can make a block (with curly braces)
anywhere I can code a statement. Are you telling me that there
/is/ a block everywhere there /can/ be a block, even though I
don't code the curly braces - or are you simply reminding me that
the potential for creating all these blocks is lurking there? Is
there block scope without a block?

 

[Ben Pfaff]

I could have, but didn't. I can make a block (with curly braces)
anywhere I can code a statement. Are you telling me that there /is/ a
block everywhere there /can/ be a block, even though I don't code the
curly braces - or are you simply reminding me that the potential for
creating all these blocks is lurking there? Is there block scope
without a block?

My guess is that C99 has every iteration statement open a block
because `for' statements in C99 can have their own local
variables (declared in the first clause).

 

[Alex]

Alex wrote:

I could have, but didn't. I can make a block (with curly braces)
anywhere I can code a statement. Are you telling me that there
/is/ a block everywhere there /can/ be a block, even though I
don't code the curly braces - or are you simply reminding me that
the potential for creating all these blocks is lurking there? Is
there block scope without a block?

I am saying that there is /block scope/ at the places which I pointed
out even though there aren't actual physical curly blocks there.

Ben Pfaff just posted a quote from the standard which seems to
support my assertion.

Alex

 

[Keith Thompson]

I could have, but didn't. I can make a block (with curly braces)
anywhere I can code a statement. Are you telling me that there /is/ a
block everywhere there /can/ be a block, even though I don't code the
curly braces - or are you simply reminding me that the potential for
creating all these blocks is lurking there? Is there block scope
without a block?

The definition of "block" changed from C90 to C99. In C90, a block is
an optional declaration list, followed by an optional statement list,
all surrounded by curly braces. In C99, not all blocks have curly
braces. So the following:

while (1) printf("I will not write infinite loops\n";

contains no blocks in C90, but two blocks (the loop and its body) in
C99.

I think the change was made because of the addition of declarations
in for loops:

for (int i = 1; i <= 10; i ++) ...

The loop was defined to be a block to provide a scope for the
declaration of i.

 

[Nick Keighley]

so by this time you should know it all

(Sam's series), and for nested loops, he writes (p116) "It's often
necessary to create a loop even when you are already in a loop." Then he
goes on to portray a contrived example that doesn't tell me under what
conditions a nested loop might be favoured as a solution? i.e. what are
nested loops useful for? What kinds of algorithms are served by nested
loops? etc. Is any of this making sense?

Anyway - thoughts welcomed.

One example you may come across in your book is a basic bubble-sort which
will sort an array of elements, e.g.

for (i=0; i<MAX-1; i++)
for (j=0; j<MAX-1-i; j++)
if (ELEMENT[j+1]>ELEMENT[j])
{
/*swap the elements*/
TEMP = ELEMENT[j+1];
ELEMENT[j+1] = ELEMENT[j];
ELEMENT[j] = TEMP;
}

Thanks Allan. I have heard about bubble sort but have not yet
encountered it. That pleasure yet awaits me!!! The code you have
listed here, would this be the standard algorithm for the bubble-sort
method, or are there also other ways of tackling this problem?

note Bubble Sort isn't a very good algorithm (there are faster methods
for
both small and large amounts of data). It is easy to code though.
For real work the standard C library comes with a standard function
qsort().
Unfortunatly qsort() comes with no performance guarantee. It might
even be Bubble sort internally! qsort() is good as a first choice;
replace it if you
know (by measurement) that it is too slow. There are books that
discuss which sort to use when.


--
Nick Keighley

"Beware of bugs in the above code;
I have only proved it correct, not tried it."
-- Donald Knuth

 

[Morris Dovey]

Thanks all. I need to spend more time reading C99, rather than
just looking stuff up from time to time.

Seems like a very strange way to rationalize "anywhere"
declarations/definitions - but perhaps it'll make more sense to
me if I read it enough times...