PirosB3

{
  "txid": "c15083d9593abad0e0ee35afa12ae7463497d34de0bc9a51e86b16634b5e9779",
  "version": 1,
  "locktime": 0,
  "vin": [
    {
      "txid": "592928d43a21294fce7592edac21060aa47cc22d9c368c544cdcdb8ef178900c",
      "vout": 0,
      "scriptSig": {
        "asm": "304402200e5ea2c9944770d5f823316565a7a1bd1131caeb203c827b47aa5d14596f80c502203e2fd2b99b39127d6f0cc5f30791048f890e2fe03cdc5d0c10e4b542b7ff8d0f01 036613cf1e844edde97f89468ce453bb609d3270a986ca9fbb6dd29d90773bf67e",

PirosB3

Hexagon: building a graph database on top of a key-value store

I've always been interested in how information is stored and retrieved. Inspired by a project of a friend (Levelgraph), I decided to research more about Hexastores and how they can be adapted to modern NoSQL databases such as LevelDB and Cassandra. Out of this fascination and curiosity I created Hexagon: a simple Python implementation of Hexastores backed by LevelDB.

In this talk I plan on giving a very simple and hands-on introduction to the subject, followed by a demonstration of how Python really allowed me to focus on the implementation thanks to its extremely intuitive syntax and features. I will show how generators, collections, and Test Driven Development allowed me to iterate fast and not get lost in (too many) bugs along the way.

The talk will start with an overview of what Hexagon is and the reason why I built it. The second step will consist in the explanation of the two main concepts behind Hexagon itself, namely LevelDB and the

PirosB3

my_request = flask.make_response('asd', 200, {'Authorization': 'bearer xxxxxxx'})

PirosB3

package main

import (
    cryptorand "crypto/rand"
    "math/rand"
    "errors"
    "sync"
    "fmt"
    "time"
)

PirosB3

1 0 0 0
0 1 0 0
0 0 1 1/f
0 0 0 0

PirosB3

[{"id":35,"user_id":36,"video_path":"6cc56475885fbdb83a536d78b16e1e0c/h264/play.mp4","description":null,"thumbnail_path":"6cc56475885fbdb83a536d78b16e1e0c/h264/play_1.jpg","created_at":"2014-01-14T21:47:10.137+01:00","updated_at":"2014-01-14T21:47:25.101+01:00","title":"jj","panda_video_id":"6cc56475885fbdb83a536d78b16e1e0c","panda_status":"complete","width":640,"height":480,"type":"video","video_url":"http://tascout.s3.amazonaws.com/6cc56475885fbdb83a536d78b16e1e0c/h264/play.mp4","thumbnail_url":"http://tascout.s3.amazonaws.com/6cc56475885fbdb83a536d78b16e1e0c/h264/play_1.jpg","likes":[],"comments":[{"id":62,"user_id":36,"user_name":"Daniel Pyrathon","profile_image_url":"http://graph.facebook.com/665044728/picture?type=large","body":"fuck off"},{"id":61,"user_id":36,"user_name":"Daniel Pyrathon","profile_image_url":"http://graph.facebook.com/665044728/picture?type=large","body":"fuck off"},{"id":60,"user_id":36,"user_name":"Daniel Pyrathon","profile_image_url":"http://graph.facebook.com/665044728/picture?ty

PirosB3

SortOfMeshSubset *LoadMesh(LPSTR filename) {
	std::wifstream		fileStream;
	std::wstring		line;
	
	// Instead of using VertexXYZ I am just using XMFLOAT3 and XMFLOAT2
	std::vector <XMFLOAT3> vectorVertices(0);
	std::vector <XMFLOAT2> vectorTextureVertices(0);
	std::vector <XMFLOAT3> vectorNormalVertices(0);

	std::vector<SimpleVertex> vertices;

PirosB3

#include <math.h>

#define WARP_SIZE 16
#define DEBUG false

/* ---------------- [[CUDA KERNELS]] ---------------- */

__global__ void updateWeightsCUDA(float *weights, float *changes, float *delta_outputs, float *inputs, int n_inputs, int n_outputs) {
    int width = n_outputs;
    int height = n_inputs;

PirosB3

import datetime
import json
import os
import time

import pymongo

from flask import Flask, render_template
from pymongo.uri_parser import parse_uri

PirosB3

import argparse
from collections import defaultdict
import cPickle, gzip
import multiprocessing
import sys

import Image
import numpy as np

from sklearn.datasets import fetch_mldata