Re: PRF & PLR flags max 32?

From: Elie Rosenblum (
Date: 01/19/02

On Sat, Jan 19, 2002 at 12:36:58PM +0100, Templar Viper wrote:
> I was lately putting in a new piece of code that required an empty pfr flag
> slot.
> However, this slot was number 32, and I later found out that it wasn't
> working
> correctly. So I now know slots like prf etc can only contain 32 flags (zero
> included).. Only, how to tackle this. Would it be more easy to create an
> entire
> new prf slot list (prf2) or is there a method I could increase the number of
> flags?
> Did anybody do this before? How did they solve it?
> If someone is willing to help me, thank you.

I decided to be a complete freak and switch to C++. I created a new
'bitvect' class, and typedefed bitvector_t to it. Internally, the
bitvect class is stored as an array of longs. Externally, the access
methods don't change much from the way the old limited bitvectors
worked, since I filled bitvect.h with inlined operator overloads to
make most of the transition completely seamless, and a few inlined
methods when it wasn't seamless. One problematic difference is I/O;
when not using asciiflags (which had to be hacked as well), I have
to use .read() and .write() methods, which need to use a FILE * or
FBFILE * directly. There's also some utility methods like .anybits(),
.inrange() (for the bpl20 DB sanity checking function), .print() (for
things like do_stat_*).

As an example, here's the bitvect.inrange() method:
class bitvect {
  bool inrange(size_t count) {
    int start;
    start = count / 32;
    count %= 32;
    if (vector[start] > (~0UL >> (sizeof(unsigned long) * 8 - count)))
      return FALSE;
    for (start++; start < MAX_BITVECT; start++)
      if (vector[start])
        return FALSE;
    return TRUE;

And here's a couple of operators, which are marked friends of the class:
inline bitvect operator |(bitvect a, bitvect b)
  bitvect tmp;
  for (int i = 0; i < MAX_BITVECT; i++)
    tmp.vector[i] = a.vector[i] | b.vector[i];
  return tmp;

inline bitvect &operator &=(bitvect &a, bitvect b)
  for (int i = 0; i < MAX_BITVECT; i++)
    a.vector[i] &= b.vector[i];
  return a;

One big but really easy to handle difference is that all of the old
bitvector #defines are gone; all PRF_* (and AFF_ and ITEM_ and every
other bit vector) flags are now actually variables (though I could
probably set them const). They are all prototyped at the bottom of
bitvect.h, and I created to have a place to instantiate
all of the vectors as actual variables and give them values.

For example, in bitvect.h:
extern bitvect nobits;
extern int FORMAT_INDENT;
extern bitvect ROOM_DARK; /* Dark                       */
extern bitvect ROOM_DEATH; /* Death trap                */
extern bitvect ROOM_NOMOB; /* MOBs not allowed          */

bitvect nobits;
bitvect ROOM_DARK(0, 0);        /* Dark                 */
bitvect ROOM_DEATH(0, 1);       /* Death trap           */
bitvect ROOM_NOMOB(0, 2);       /* MOBs not allowed             */

My constructor takes two numbers, one an index into the array of
vectors and the other a small integer for bit shift value. It would
be easy enough to just take one integer instead that you would div
by your long size and mod to get the bit shift.

It's by no means a quick and easy transition, but it worked. I
would not recommend it for the faint of heart or weak of code though.

What might work for you is to redefine your bitvectors to 'unsigned
long long'. You will have to change some code but depending on your
compiler that might yield 64-bit vectors.

Is anyone actually interested in my bitvect.h and I
would be happy to distribute those, but the changes that happen
elsewhere in the code are too widespread for me to make a patch, and
I don't want to hand out my entire source. :) If anyone wants to
attempt it, I'd be happy to help out with questions about how to
handle transitioning pieces of code (I've currently got it working
with OLC, ascii pfiles, and ascii rent files, so it's definitely

Elie Rosenblum                 That is not dead which can eternal lie,   And with strange aeons even death may die.
Admin / Mercenary / System Programmer             - _The Necronomicon_

   | FAQ: |
   | Archives: |
   | Newbie List:   |

This archive was generated by hypermail 2b30 : 06/25/03 PDT