Calculating the cardinality of a set is an easy task, but
is there any Delphi BUILT-IN function doing the job?
I looked at the Delphi Help and at developers.href.com
site and found nothing about such a function.
Thanks in advance

Jacques Melancon wrote:
  (snip)

What do you mean with the cardinality of a set?
  --
  Rudy Velthuis
  

a set is a predefined type, just count them.
  Rene
  --
  Ing.Buero R.Tschaggelar - http://www.ibrtses.com
  Jacques Melancon wrote:

I know I can count. I want to know if there is a
    BUILT-IN function doing the job for me.
    Jacques
    Rene Tschaggelar wrote:
    (snip)

<       I know I can count. I want to know if there is a
      BUILT-IN function doing the job for me.
      (snip)
There is no built-in function and, like you, I think there should be, just
      as there should be sets beyond 256 elements. Many years ago, I wrote a

Philippe Ranger wrote:
        (snip)

Weren't we just here?
        Your "many years ago" is in full flower. That 'loop' instruction has
        got to go. Besides, it would seem that wrapping around the best of the
        "bitsInByte" solutions would be both faster and less cumbersome.

Merci Philippe,
        As you say I think it would be usefull if we could have such a function.
        Now, a question about your function. If I declare something like:
        Type
        TDigits = set of 0..9;

<           I would call your function like this:
          Cardin := Card( Digits, sizeof(TDigits) );
          or
          Cardin := Card( Digits, 2 );

Philippe Ranger <.> wrote in message news:7u55ru$✉forums.borland.com...
            (snip)

Philippe
            1. I know that there is no bit addressing
            2. Both calls are identical in my example ( sizeof(TDigits) IS EQUAL TO 2 ).
            You need at least 2 bytes to store 10 "boolean" values.

<               1. I know that there is no bit addressing
              2. Both calls are identical in my example ( sizeof(TDigits) IS EQUAL TO 2 ).
              You need at least 2 bytes to store 10 "boolean" values.
              (snip)
Sizeof() is not a function, Jacques, it's a compiler macro.

The cardinality of a finite set is the number of elements it has.
    Phillipe has a BASM solution to this problem that he's posted a couple
    of times. Maybe he'll do it again.
    Glynn
    Rudy Velthuis wrote:

Wouldn't that be High(set)???
      Glynn Owen <✉gtstrans.com> wrote in message
      news:✉gtstrans.com...
      (snip)


      

Eddie Shipman wrote:
        (snip)
That would be the maximum number of items a set can have. The
        cardinality would be the number of items actually in there.
        The answer is no, there are no built-in functions for this. If
        Philippe posted code, try to find it on

Shouldn't that be like ord(high(set)) + 1 ?
        
        
        
        

<           Shouldn't that be like ord(high(set)) + 1 ?
          (snip)
Not at all sure. What happens with a set of 14..28?
          PhR
          

Philippe, You are are quite correct in seeing a problem with
            "ord(high(set))+1". On a related topic, that I have never seen
            discussed here, is the problem of using other people's definition of
            enumerated types in const statements like ...
            const FieldTypeNames: array [TFieldType] of string[4] = ('Unk', 'A',

<               But what is a good way to minimize the effect of adding a member in
              the middle of the range?
              (snip)
Far as I know, none. Whoever does this on a commercial library is a criminal
              (unless the type maps an OS-defined thing, which changed).

<     What do you mean with the cardinality of a set?
    (snip)
Guess why my bit counter is named Card.
    PhR
    

<           Besides, it would seem that wrapping around the best of the
          "bitsInByte" solutions would be both faster and less cumbersome.
          (snip)
Perhaps, but that involves one call per byte. Feel free to redo Card if you
          want.

Philippe Ranger wrote:
            (snip)

By wrap I meant put the code inside a loop. This is between 1.7 and 3x
            faster depending on the size of the set on a PII, and it uses your
            BitsInByte Nibble algorithm crushed into one pascal expression.
            function card1(var tar; size: integer): integer;

<               By wrap I meant put the code inside a loop. This is between 1.7 and 3x
              faster depending on the size of the set on a PII, and it uses your
              BitsInByte Nibble algorithm crushed into one pascal expression.
              (snip)
This is getting embarrassing!

Philippe Ranger wrote:
                (snip)

Actually, I didn't 'work' on it. I just applied my standard tune-ups,
                then looked at what I got (CPU window). It looked good so I quit.
                (snip)
For loops usually have more setup overhead and usually increment a
                counter and one pointer for each array in the loop. The while loop has

<                   I don't. I scan up referened from the back of the array. Thus i goes
                  from -size to -1.
                  (snip)
Got it!
                  <<

Philippe Ranger wrote:
                    (snip)

Absolute implies memory address. However, many params don't have any
                    memory associated with them as they are values. By using the 'absolute'
                    approach to typecasting them you force them on to the stack and out of a
                    register, making them slower in the process. Upon reflection though,

<<                       From way back, I see absolute as the normal complement for untyped params.
                      Can't say your pointer and assignment look like a convincing improvement.
                      (snip)
<                       Absolute implies memory address. However, many params don't have any

Philippe Ranger wrote:
                        (snip)

The while loop really isn't optimized at all in the way for loops are.
                        A while loop is handled more like a bent if-then statement.
                        Consequently, you can manually optimize access within the loop by proper
                        use of pointers, indicies etc. But you have to do it all yourself. I