nborwankar

So what exactly is in a Sproingg node?
-------------------------------------

Sproingg is a project to bootstrap a richly featured text communication system over CouchDB using 
the following apps and CouchDB features collectively referred to as a Sproingg node

a) A CouchDB database as message inbox with a CouchDB database URL as an address for sending messages (Reqd)
b) A CouchApp that is a forum app (Optional)
c) A CouchApp that is a wiki app (Optional)
d) A CouchApp blog (Optional)

nborwankar

testing ....

nborwankar

THE PROBLEM:

Standard practices say no non-root process gets to talk to the Internet on a port less than 1024.  How, then, could I get Node talking on port 80 on EC2?  (I wanted it to go as fast as possible and use the smallest possible share of my teeny tiny little micro-instance's resources, so proxying through nginx or Apache seemed suboptimal.)  


THE TEMPTINGLY EASY BUT TOTALLY WRONG SOLUTION:

Alter the port the script talks to from 8000 to 80:

}).listen(80);

nborwankar

curl -X PUT http://your:credentials@localhost:5984/profiles

curl -X POST -d '{"source":"http://couchcamp.couchone.com/profiles","target":"http://localhost:5984/profiles","continuous":true}' -H "Content-type: application/json" http://localhost:5984/_replicate

curl -X POST -d '{"target":"http://couchcamp.couchone.com/profiles","source":"http://localhost:5984/profiles","continuous":true}' -H "Content-type: application/json" http://localhost:5984/_replicate

nborwankar

#!/bin/bash

#
# This script installs and configures couchdb on a fresh Amazon Linux AMI instance.
#
# Must be run with root privileges
# Tested with Amazon Linux AMI release 2011.02.1.1 (ami-8c1fece5)
#

export BUILD_DIR="$PWD"

nborwankar

Here's an R session mimicking the steps in the doc for decision trees.  At this point not meant to show anything like the final result just a probe of the R<->Pg semantics.
Note that the driver seems to create a result set in a format that the data frame constructor in R understands.

Not only that but the equivalent of "to_string" for df seems to output it in the familiar tabular format by default.
So the happy result to report here is that the native impedance mismatch is small.
We'll need some validation tests to identify R<-->GP issues if any.

Right now I am using an R<-->Postgres driver (MADlib 0.4 on Pg 9.1 on MacOSX Lion., RStudio 0.9.4, R 2.13.2).

nborwankar

===

> rs <- dbSendQuery(con, statement = paste( 
+ "SELECT * FROM",		+"MADlib.c45_train('infogain','public.golf_data','trained_tree_infogain',null,'temperature,humidity','outlook,temperature,humidity,windy','id','class',100,'explicit',	5,0.001,0.001,0)"));

> df <- fetch(rs, n = -1)
> df
 		 training_set_size tree_nodes tree_depth   training_time split_criterion
		                14          8          3   00:00:00.311362        infogain

nborwankar

# compute error measures for predictive model for activity (act) as dependent variable
# note this is computed for all 6 activities and a total computed which is an aggregate value

errormeasures <- function(orig, pred, act) {
    N <- length(orig)
    origtrue  <-as.vector(NULL)
    origfalse <-as.vector(NULL)
    predtrue  <-as.vector(NULL)
    predfalse <-as.vector(NULL)
    

nborwankar

# pgdb.py

__author__ = 'nitin'
import psycopg2 as pspg2
import sys
# import pandas as pd, then use pandas.io.sql

db_data = {
    'host':'localhost',
    'dbname': 'nitin',

nborwankar

__author__ = 'nitin'

# -*- coding: utf-8 -*-
"""
===========
pgmagic
===========

Magics for interacting with Postgres database server via psycopg2.