short first draft readme
read COPYING, the license for this code is GPL
If you end up using this for any purpose, please drop me a mail. The code was written for my own uses, but I spent a day getting it prepped for release so I'd love to know if it was worth it. (I'll remove that paragraph if I get a few emails)
- INSTALL
- python setup.py install
This will put it in your python site-packages/ directory, wherever that is
This source is currently only easilly used as a Python module I intend to library-ize the C in the future, but it isn't there yet I intend to write perl bindings " " " " " " " "
Here is a short sample Python program
import probstat
mylist = [1,2,3]
p = probstat.Permutation(mylist)
for (item) in p:
print item
generates the output
[3, 2, 1]
[2, 3, 1]
[3, 1, 2]
[1, 3, 2]
[2, 1, 3]
[1, 2, 3]
which is all the permutations of 1,2,3
you can do random access
l = p[17] # might throw an IndexError though
and slices
l = p[1:7] # possible IndexError
and of course length
print len(p)
DEPRECATED, the 'next()' method is deprecated, use for (blah) in foo:
_PATCHES WELCOME!_
The C code should be very readable, hack away. All the code passes ElectricFence's memory checks, but if you find an error please mail me, or prefereably submit a patch
As for performance, it is 10x + faster than the pure-python version
(which I might post if people want it)
Memory consuption is more than it needs to be by a factor of about two
this is because I wrote the C version to know nothing about the python bindings
Since the python bindings know everything about the C code, I could eliminate
this. I'll get around to it someday.
Unlike some of the perl modules that do a similar thing, we don't calculate all the premutations/combos/etc at the beginning and then iterate over the [possibly] huge list. When you iterate it figures out what the next item is. The memory consumed is about the size of the list passed in times three times the size of a pointer (we don't copy the list, just pointers to the items in the list).
If the list is straight scalars (int, string, etc) you can change the orignal list and not worry about the results of the Permutation changing. I only keep a shallow copy of the data, so if you are permuting a list of lists and you change the second-tier lists I make no garuntees that the data out is what you want.
email jackdied@users.sourceforge.net with ideas, patches, comments (if that bounces try jack_diederich@email.com)