I have been using python for Civilization IV map scripting, and I 
believe I have found a small problem with the randint function. Once in 
a blue moon, randint(a,b) returns b + 1, rather than a >= N <= b. I 
have seen this behavior a few times. Here is the code I used to 
determine this.

randIndex = PWRand.randint(0,len(preshuffle)-1)#PWRand calls randint 
directly with these parameters
if randIndex < 0 or randIndex >= len(preshuffle):
    raise ValueError, " bad index shuffling plot list randIndex=%(r)d 
listLength=%(l)d, mapSize=%(s)d" % {"r":randIndex,"l":len
(preshuffle),"s":hm.mapWidth*hm.mapHeight}

#output is
ValueError:  bad index shuffling plot list randIndex=1453 
listLength=1453, mapSize=13824

according to the docs, the max randIndex should be listLength - 1

I have been using python for Civilization IV map scripting, and I 
believe I have found a small problem with the randint function. Once in 
a blue moon, randint(a,b) returns b + 1, rather than a <= N <= b. I 
have seen this behavior a few times. Here is the code I used to 
determine this.

randIndex = PWRand.randint(0,len(preshuffle)-1)#PWRand calls randint 
directly with these parameters
if randIndex < 0 or randIndex >= len(preshuffle):
    raise ValueError, " bad index shuffling plot list randIndex=%(r)d 
listLength=%(l)d, mapSize=%(s)d" % {"r":randIndex,"l":len
(preshuffle),"s":hm.mapWidth*hm.mapHeight}

#output is
ValueError:  bad index shuffling plot list randIndex=1453 
listLength=1453, mapSize=13824

according to the docs, the max randIndex should be listLength - 1

I've done some more testing and I can't get this to repeat with randint 
alone. It must be an issue with combination of the len() function used 
as a parameter to randint. The above code is in a loop that further 
down will del preshuffle[randIndex]. Could this be a threading issue 
where len(preshuffle) takes time to update when the list is shortened?

Threading is the likely source of your problems.  If you concur, please 
close this as invalid.

To be clear, I am not using multi-threading in my particular module. I 
can't explain this behavior with my code alone. The issue is 
complicated by the fact that my module is called by Civ IV, and I have 
no idea what the game is doing behind the scenes. I also don't know 
exactly what happens when 'del myList[index]' is called. You would 
think that len(preshuffle) would have the same value for the randint 
call that it does for the print statement. I have an idea for another 
test I would like to try.

This is my first issue that I've reported in Python. Would it help for 
me to attach the whole file?

Attaching the whole file isn't a step in the right direction.  The 
preferred approach is to isolate the problem as tightly as possible.

This report is dubious because, I can't get the following to fail:

from random import seed, randint
seed('mystart')
while 1:
   assert 0 <= randint(0, 1452) <= 1452

There is also no failure with:

preshuffle = [None] * 1453
while 1:
   assert 0 <= randint(0, len(preshuffle)-1) <= 1452

Make sure that PWRand.randint() really does call randint() and not 
randrange() and that it doesn't adjust the range of inputs.

What was happening before was I was getting an index out of range error 
every so often, so then I put in the ValueError catch to see what was 
going on. I was surprised to see that randIndex was the same as len
(preshuffle). I have some further catches in place now that might tell 
me more. It must have to do with deleting items from the list in the 
loop and not getting the right length from len immediately after, but 
then again I use this algorithm in several places in my module with no 
trouble, though with smaller lists. Go ahead and close this, and when I 
have more information I can open it some other time. This bug I'm 
seeing is extremely intermittant.