[CF-Devel] Improvement needed on artifact selection code?

[Peter Mardahl]

Well, I tried generating roughly 1000 rings at random,
all artifacts.

The numbers which showed up were all within expectation
for a random distribution of picks, except:

1)  Too many Rings of War showed up
   (not really significant since their numbers were low:
    I got six, but expected 3)

2)  Too few rings of ruling showed up
    (1 instead of 3)

However, neither 1) nor 2) is probably significant due to
the small sample involved.

PM

>
     
      Peter Mardahl wrote:
     
     >
     
      > 
     
     >
     
      > Hello,
     
     >
     
      > 
     
     >
     
      >   I recently added a bunch of rings oriented for fighters,
     
     >
     
      > which grant fighting stats at the expense of magic abilities.
     
     >
     
      > 
     
     >
     
      >   These rings are popular among the masses, and unfortunately,
     
     >
     
      > they're common.
     
     >
     
      > 
     
     >
     
      >   For some reason, a Ring of Thieves, for example, comes up
     
     >
     
      > more often than a Ring of Halvor.
     
     >
     
      > 
     
     >
     
      > Thieves has a chance of 20
     
     >
     
      > Halvor has a chance of 50
     
     >
     
      > 
     
     >
     
      > Therefore, you should see 2.5 rings of halvor for every ring
     
     >
     
      > of thieves.  Instead, you see more rings of thieves.
     
     >
     
      > 
     
     >
     
      > >From my vague recollections of the code, items up at the top
     
     >
     
      > are favored in the artifact selection over items at the bottom.
     
     >
     
      > 
     
     >
     
      > Is this true?  Or is the observation just a result of chance?
     
     >
     
     
     >
     
       At least looking at the code, what you describe should be correct (See 2.5
     
     >
     
      rings of halvor for every ring of thieves).
     
     >
     
     
     >
     
       The selection of an artifact does not depending on the ordering.  After load
     
     >
     
      time, the chance values for all the objects of a type are summed up and store
     
     >
     
     d. 
     
     >
     
      When it generates a random artifact, it generates a random number from 0 to t
     
     >
     
     hat
     
     >
     
      chance value, and traverses the list until the appropriate place is found
     
     >
     
      (subtracts item chance from roll, if less than 0, use that artifact, otherwis
     
     >
     
     e
     
     >
     
      continue with same roll)
     
     >
     
     
     >
     
       Now there may be some bug someplace - perhaps it isn't totalling the chance
     
     >
     
      right or doing something else odd.  But order in the artifact file should not
     
     >
     
      make a difference.
     
     >
     
      _______________________________________________
     
     >
     
      crossfire-devel mailing list
     
     >
     
     
      crossfire-devel at lists.real-time.com
      
      
     >
     
     
      https://mailman.real-time.com/mailman/listinfo/crossfire-devel