Noticed some odd behavior during play testing, finally tracked it back to this bit.

So if you're in the one_hit( ) function and you attempt to damage someone, say 30 points of blunt damage, you eventually land here:

   if( ( retcode = damage( ch, victim, dam, dt ) ) != rNONE )
      return retcode;
      
   if( char_died( ch ) )
      return rCHAR_DIED;
      
   if( char_died( victim ) )
      return rVICT_DIED;

   retcode = rNONE;
   
   if( dam == 0 )
      return retcode;

   /*
    * Weapon spell support            -Thoric
    * Each successful hit casts a spell
    */
   if( wield && !IS_SET( victim->immune, RIS_MAGIC ) && !xIS_SET( victim->in_room->room_flags, ROOM_NO_MAGIC ) )
   {
    
   }

And so on. The problem is that damage( ) might not actually inflict any damage, seeing as how one might be immune to, in this case, blunt damage. However, one_hit( ) still has a positive 'dam' value, so you won't bail out with the dam == 0 check above, and hence you'll end up with weaponspells still firing on what I think most would consider an unsuccessful hit.

A simple fix might be to check the HP of the victim before and after the damage call. No change would be a sign of a unsuccessful hit, so you could do something like this:

    previous_hp = victim->hit;
    
    if ( (retcode = damage( ch, victim, dam, dt )) != rNONE )
    	return retcode;
    	
    if ( char_died(ch) )
    	return rCHAR_DIED;
    	
    if ( char_died(victim) )
    	return rVICT_DIED;

    retcode = rNONE;

    if ( dam == 0 )
	    return retcode;

   /*
    * Weapon spell support            -Thoric
    * Each successful hit casts a spell
    */
   if( wield && !IS_SET( victim->immune, RIS_MAGIC ) && !xIS_SET( victim->in_room->room_flags, ROOM_NO_MAGIC ) && victim->hit < previous_hp )
   {
    
   }

This works for me, and preserves the functionality of the retaliatory shields, which makes intuitive sense (e.g. still close enough to get burned by a fire shield, even if your hit didn't do any damage). Obviously don't forget to declare the variable somewhere, up top in one_hit( ) or whatever suits you.

Perhaps this is one of those items open to some measure of debate, but given the way the code was written, I don't see why immunity to PLUSx, NONMAGIC and MAGIC (checked in one_hit( ) ) should be any different from immunity to everything else, i.e. BLUNT (checked in damage( ) ), in that all should be considered unsuccessful hits if damage = 0.

Thoughts?