teh

{-# LANGUAGE ScopedTypeVariables #-}
{-# LANGUAGE TypeApplications #-}
{-# LANGUAGE AllowAmbiguousTypes #-}

{-# LANGUAGE DataKinds, PolyKinds, TypeOperators #-}
{-# LANGUAGE TypeFamilies, FlexibleInstances #-}
{-# LANGUAGE InstanceSigs #-}

-- When do I need TypeApplication?
-- When I need some intermediate types that change how

teh

#!/usr/bin/env python
"""
Welcome to Tom-does-another-disgusting-hack.

This is an executable shim for the stack binary whose only purpose is
to make the amazing intero mode [1] work in a non-stack environment (Nix
in our case).

Dump it somewhere and make sure it runs *instead* of the actual stack binary.
You will not be able to use stack at the same time.

teh

import sklearn.base as base
import sklearn.linear_model as lm
import numpy


class ByThreshold(base.BaseEstimator, base.ClassifierMixin):
    def __init__(self, estimator, threshold=0.95):
        self.threshold = threshold
        self.estimator = estimator

teh

#!/usr/bin/env python

import sys
import time
import os
import subprocess

def main():
    if sys.argv[1:] == ["path", "--project-root", "--verbosity", "silent"]:
        print "/home/tom/src/intero"

teh

$ cat .git/config 
...
[diff "nixops"]
	textconv = nixops export -s

$ cat .gitattributes 
production/secrets/**/* filter=git-crypt diff=nixops

teh

{%- extends 'full.tpl' -%}

{%- block input_group -%}
{%- endblock input_group -%}

teh

{ fetchurl, stdenv }:

stdenv.mkDerivation rec {
  name = "pies-1.2";

  src = fetchurl {
    url = "mirror://gnu/pies/${name}.tar.bz2";
    sha256 = "18w0dbg77i56cx1bwa789w0qi3l4xkkbascxcv2b6gbm0zmjg1g6";
  };

teh

import numpy
import timeit
import numba

@numba.autojit(locals=dict(
    best_k=numba.int64,
    pos=numba.int64,
    ins_del_count=numba.int64,
    i=numba.int64,
    j=numba.int64,

teh

# Solve n-queens problem with picosat
import pycosat
import numpy
import itertools

def get_cnf(N):
    cnf = []

    # * convert to object because pycosat expects 'int's
    # * add 1 because 0 is reserved in pycosat

teh

from PIL import Image
import numpy
from skimage.filter import sobel
from skimage.morphology import watershed
from scipy import ndimage as nd

grind = numpy.asarray(Image.open('grind.png')).mean(axis=2)

edges = sobel(grind)
markers = numpy.zeros_like(grind)