radix

## IO Monad for Python

Effects are vaguely analogus to IO monads. The Effect class can be compared
to the IO type, which tags (or wraps) your result type, and
`Effect.on_success` is somewhat like the bind function (`>>=`), indicating
that the function passed is to be called with the result of the effect.
Haskell's "Either" can be thrown in to handle `.on_success` vs `.on_error`.

But Effect is a little more than just the IO monad, since Effects make
available the intent as *transparent data*. By transparent, I specifically

radix

def get_request(method, url, **kwargs):
    """Return a Request wrapped in an Effect."""
    return Effect(Request(method=method, url=url, **kwargs))


def request_with_reauth(get_request, method, url, auth=None,
                        headers=None, **kwargs):
    """Create a request which will reauthenticate upon 401 and retry."""

    def handle_reauth(result, retries):

radix

def wrap(f, g):
    """
    'The wrapper combinator'

    Given f, any function, and g, a function which accepts a callable as its
    first argument, return a function:

        λ(*args, **kwargs): g(f, *args, **kwargs)

    This allows g to "wrap" f, so that g is responsible for calling f. f must expect a

radix

radix@伝説:~$ python3
Error processing line 1 of /Users/radix/Library/Python/2.7/lib/python/site-packages/virtualenvwrapper-4.2-py2.7-nspkg.pth:

  Traceback (most recent call last):
    File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site.py", line 167, in addpackage
      exec(line)
    File "<string>", line 1, in <module>
  ImportError: No module named 'new'

Remainder of file ignored

radix

"""
An immutable bag/multiset implementation for Python.
"""

import six

if six.PY2:
    from collections import Container, Iterable, Sized, Hashable
else:
    from collections.abc import Container, Iterable, Sized, Hashable

radix

def _converge_lb_state(desired_lb_state, current_lb_state, ip_address):
    desired = {
        (lb_id, config.port): config
        for lb_id, configs in desired_lb_state.items()
        for config in configs}
    current = {
        (node.lb_id, node.config.port): node
        for node in current_lb_state}
    desired_idports = set(desired)
    current_idports = set(current)

radix

from __future__ import print_function

from characteristic import attributes
from effect import Effect, perform, ConstantIntent, base_dispatcher, sync_performer
from effect.dispatcher import TypeDispatcher, ComposedDispatcher


@attributes(['name'])
class MyThing(object):
    """Intent."""

radix

def atom(a):
    if a == 0: return -1
    else: return a

def composition(a):
    return ('composed', atom(a))

radix

"""
Re-raising with tracebacks in Python is confusing.

First, I was surprised that when you re-raise with an explicitly specified
traceback, you don't get that exact traceback from sys.exc_info() after
catching it. Then I realized Python adds more information upon
re-raising. Okay, that's great!

But then I noticed that the re-raising frame is left out of the traceback!
That's *super* confusing! Note how in the "new" traceback, there's a frame in

radix

from __future__ import print_function

import copy

from pyrsistent import PRecord, field

NUM_FIELDS = 20


def make_pfoo(attr_names):